Вы находитесь на странице: 1из 62

NBER WORKING PAPER SERIES

CORPORATE TAXES AND INTERNAL BORROWING WITHIN MULTINATIONAL


FIRMS
Peter Egger
Christian Keuschnigg
Valeria Merlo
Georg Wamser
Working Paper 18415
http://www.nber.org/papers/w18415
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
September 2012
Peter Egger has not received any external funding for this project. Christian Keuschnigg has not received
any external funding for this project. Valeria Merlo has not received any external funding for this project.
Georg Wamser has not received any external funding for this project. The views expressed herein
are those of the authors and do not necessarily reflect the views of the National Bureau of Economic
Research.
NBER working papers are circulated for discussion and comment purposes. They have not been peer-
reviewed or been subject to the review by the NBER Board of Directors that accompanies official
NBER publications.
2012 by Peter Egger, Christian Keuschnigg, Valeria Merlo, and Georg Wamser. All rights reserved.
Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided
that full credit, including notice, is given to the source.
Corporate Taxes and Internal Borrowing within Multinational Firms
Peter Egger, Christian Keuschnigg, Valeria Merlo, and Georg Wamser
NBER Working Paper No. 18415
September 2012
JEL No. F23,G32,H25
ABSTRACT
This paper develops a theoretical model of multinational firms with an internal capital market. Main
reasons for the emergence of such a market are tax avoidance through debt shifting and the existence
of institutional weaknesses and financial frictions across host countries. The model serves to derive
hypotheses regarding the role of local versus foreign characteristics such as profit tax rates, lack of
institutional quality, financial underdevelopment, and productivity for internal debt at the level of a
given foreign affiliate. The paper assesses hypotheses in a panel data-set covering the universe of German
multinational firms and their internal borrowing. Numerous novel insights are gained. For instance,
the tax-sensitivity found in this paper is many times higher than previous research suggests. This accrues
mainly to three things: the consideration of the boundedness of the internal debt ratio as a dependent
variable in comparison to its treatment as an unbounded variable in most of the previous work; the
coverage of all (small and large) multinationals here rather than a focus on large units in previous work;
and the inclusion of endogenous characteristics in other countries multinationals are invested in (due
to endogenous weights) while previous work did not consider such effects at all or assumed them to
be exogenous. Moreover, local and foreign (at other locations of a given affiliate) market conditions
matter more or less symmetrically and in the opposite direction. There is a nonlinear trade-off between
institutional quality or financial development on the one hand and higher profit tax rates on the other
hand, and the strength of this trade-off depends on the characteristics of one location relative to the
other ones a multinational firm has affiliates (or the headquarters) in.
Peter Egger
ETH Zurich
Department of Management,
Technology, and Economics
Weinbergstrasse 35
8092 Zurich, Switzerland
egger@kof.ethz.ch
Christian Keuschnigg
University of St.Gallen, FGN-HSG
Varnbuelstrasse 19
CH-9000 St. Gallen
Switzerland
Christian.Keuschnigg@unisg.ch
Valeria Merlo
ETH Zurich
Department of Management,
Technology, and Economics
Weinbergstrasse 35
8092 Zurich, Switzerland
merlo@kof.ethz.ch
Georg Wamser
ETH Zurich
Department of Management,
Technology and Economics
Weinbergstrasse 35
8092 Zurich, Switzerland
wamser@kof.ethz.ch
1 Introduction
... in a credit-constrained setting [...] headquarters can create value by actively reallocating
scarce funds across projects. For example, the cash ow generated by one divisions activity
may be taken and spent on investment in another division, where the returns are higher.
Jeremy Stein (1997, p. 111)
A recent literature in corporate nance explores how large corporations allocate scarce
funds to dierent company divisions through an internal capital market (see Gertner and
Scharfstein, 1994; Stein, 1997; Bolton and Scharfstein, 1998; Scharfstein and Stein, 2000).
A prime example of such multi-unit companies are multinational enterprises (MNEs) which
operate aliated entities in dierent countries. In principal, each plant could operate as a
separate unit and independently raise external funds to nance investment, just like any other
local rm. Similar to other forms of multi-unit rms, productivity may dier across the units
(aliates) within MNEs. However, in contrast to domestic (single-country) multi-unit rms,
MNEs are exposed to dierent corporate tax rates, institutional quality, or levels of nancial
development. Unit-specic productivity and local corporate tax rates determine the return
on capital across units. Local institutional quality (such as legal, accounting, or governance
standards) aect the accessibility of external capital markets. Similarly, the maturity of
the nancial sector (or its development) inuence the loan rates charged at external capital
markets and, hence, the cost of capital. In case of scarce resources, dierences in these
fundamentals (even in the absence of tax dierentials) across countries and locations of
units of an MNE then generate an incentive to reallocate capital across units to their most
ecient use through an internal capital market. An important consequence of this procedure
is that investments of dierent units within the same MNE then become interrelated.
This paper postulates a theoretical model of aliates of MNEs which are nancially
constrained to dierent degrees. All units raise external funds at local capital markets and,
in addition, borrow or lend at the MNEs internal capital market. The presence of nancial
constraints entails an excess return on investment. In our model, tighter constraints imply
higher returns on investment. We illustrate as to how an MNE optimally allocates internal
funds among units facing lesser constraints and ones with tighter constraints through an
internal capital market. This internal borrowing and lending renders investment of units of
an MNE dierent from stand-alone, but otherwise comparable, local rms which cannot rely
on an internal capital market. By ignoring reasons for an internal capital market beyond the
variability of prot tax rates, earlier research on the matter comes to the strong hypothesis
that only the aliate facing the lowest tax rate should lend and all others borrow internally
to exploit the tax advantage of interest deductions.
The model proposed in this paper predicts a much richer pattern of internal capital ows
that cannot be explained by standard models without capital market frictions. Dierences in
unit-specic productivity, local institutional quality, and nancial institutions may amplify
or oset dierences in prot tax rates so that internal capital can ow in any direction.
Empirical work ought to control for such fundamentals of the internal capital market to
be able to identify the eect of corporate tax rates on internal borrowing and lending.
2
Furthermore, since an operative internal capital market renders investments across plants
inherently interdependent, not only the lowest prot tax rate but the tax rates of all units
are relevant.
We utilize the panel data-set Microdatabase Direct Investment (MiDi) provided by
Deutsche Bundesbank (the German Central Bank) containing information about virtually
all foreign aliates of German MNEs to shed light on both aforementioned aspects of inter-
nal capital markets: the dependence of the internal capital market on fundamentals beyond
tax rates such as institutional quality and nancial development at the location of units;
and the relevance of prot tax rates, productivity, and (nancial and other) institutions not
only at the unit with the most favorable environment but also at other locations for internal
borrowing and lending.
Using a fractional response model, our main results imply that a one-percentage point
higher statutory tax rate in the host country is associated with a 0.92-percentage point
higher internal-debt-to-capital ratio of the borrowing aliate. At the same time, a one-
percentage point increase in the (weighted) tax rate of the lending aliates is associated
with a 0.77-percentage point lower internal-debt-to-capital ratio of the borrowing aliate.
Other determinants of internal debt show the same qualitative pattern. While nancial
underdevelopment in the host country is positively related to the internal debt ratio of the
borrowing aliate, nancial underdevelopment at other locations exerts a negative eect on
the internal debt ratio of the borrowing aliate. Financial weakness in the host country is
associated with a higher internal debt ratio of the borrowing aliate, but nancial weakness
at the lender locations leads to a lower internal debt ratio of the borrowing aliate. A
higher aliate-level productivity of the borrowing aliate relates positively to its internal
debt ratio, but a higher productivity of the lenders exerts a negative impact on the internal
debt ratio of the borrowing aliate (it should be mentioned, however, that productivity
measures are found to be statistically insignicant).
The next section sets out to portray the state of earlier theoretical and empirical work on
internal capital markets and MNEs. Sections 3 and 4 introduce and analyze, respectively, the
theoretical model. Section 5 describes the empirical approach taken and introduces the data.
Section 6 summarizes and discusses the empirical results, and the last section concludes.
2 Previous Work on Debt Shifting
A substantial literature in public economics has investigated theoretically and empirically
how tax rate dierences induce international prot shifting via internal debt and other
channels. MNEs with aliates in dierent countries tend to exploit dierences in tax rates to
reduce consolidated tax payments. Devereux (2007) provides an overview of the literature. In
Mintz and Smart (2004), for example, the subsidiary in the country with the lowest tax rate
lends to other units subject to higher tax rates. This tax arbitrage is protable since interest
earnings in the tax haven country are only lightly taxed, while interest deductions in high-tax
countries create large tax savings. Typically, the literature assumes reduced form agency
costs of debt such that a deviation from a natural leverage ratio and a positive internal debt
ratio create rising deadweight costs that eventually limit the amount of debt shifting (see
3
Fuest and Hemmelgarn, 2005; Huizinga et al. 2008; Buettner et al., 2009; Overesch, 2009;
Egger et al., 2010; for example). Empirical results nd that debt shifting is important, and
undermines corporate tax revenue in high-tax countries (Huizinga et al., 2008) since MNE
subsidiaries are much higher leveraged than autonomous national rms (Egger et al., 2010).
On the positive side, the possibility of debt shifting can facilitate investment in high-tax
countries (Overesch, 2009).
Desai et al. (2004) estimate that 10% higher local corporate tax rates are associated
with 2.8% higher aliate debt-to-asset ratios. One interesting result of their analysis is
that the elasticity of external borrowing with respect to the tax rate is 0.19, while the
elasticity of internal borrowing is 0.35. Mintz and Smart (2004) argue that income shifting
has pronounced eects on provincial tax bases in Canada. They estimate the elasticity of
taxable income with respect to tax rates for income shifting rms to be 4.9 and only 2.3 for
other, comparable rms. Egger et al. (2010) nd a substantial dierence in the debt-to-asset
ratios between foreign and domestically owned rms and report a strong interaction eect
between plant operation mode and tax rates. An increase in the statutory corporate tax
rate by one percentage point leads to an increase in the debt ratio by about 0.7 percentage
points.
Some papers recognize more interaction of internal lending among aliate units, as op-
posed to the unidirectional ow of debt from the unit with the lowest tax rate to all others
with higher rates. Huizinga et al. (2008) nd that debt leverage of a subsidiary in a given
country signicantly depends on a weighted average of tax rate dierentials to all other units.
Desai et al. (2004) point to the fact that internal debt not only responds to tax rate dier-
entials but is also importantly inuenced by nancial market development and institutional
quality. They nd that multinational aliates obtain less external debt in countries with
underdeveloped capital markets and higher local borrowing costs. Internal debt substitutes
for about three quarters of reduced external borrowing due to adverse local capital market
conditions. Buettner et al. (2009) conrm this result using data on German multinationals.
Antras et al. (2009, p. 1208) argue, as does this paper, that the exploitation of technology
is central to understanding MNE activity, but the critical constraint is the nature of capital
market development and investor protections in host countries. Their focus is, however, on
the choice of arms length trading relative to foreign direct investment (FDI), rather than
tax-induced internal debt shifting. Like the present paper, they develop an MNE model
with credit rationing of local production units in the spirit of Holmstrom and Tirole (1997)
and explain how institutional and capital market variables determine the parents decision
to acquire a larger equity stake and thereby co-nance the local production unit rather than
choosing an arms length relationship. Their empirical analysis conrms that weak investor
protection and adverse local capital market conditions in the host country limit the scale of
MNE activity.
The recent empirical literature thus leads us to believe that MNEs operate internal cap-
ital markets not only to exploit tax rate dierentials across countries but, perhaps more
importantly, to overcome institutional and nancial investment barriers in host countries.
Our analysis of internal capital markets implies that rms tend to withdraw internal funds
from less protable units and reallocate them to operations with an above-average return on
investment but limited access to external nancing. Compared with the existing literature
4
on internal debt shifting, our analysis yields two central and novel results. First, investment
in dierent units tend to be inherently interrelated. Second, the tax motive may reinforce
or oset the economic motive of using internal debt, leaving no clear-cut pattern of inter-
nal debt ows. In spite of the tax disadvantage, internal funds may easily ow to high-tax
countries for good economic reasons.
Similarly, albeit in a dierent context, Lamont (1997) provides evidence that investments
are interrelated among dierent aliates: when major oil companies cash ows were hard-
hit by the oil price decline of 1986, leaving them with much reduced own funds, they cut
investment across the board, both in oil and non-oil related divisions.
1
Gopalan et al. (2007)
argue that intragroup loans are typically used to support nancially weaker rms, e.g., to
avoid default within the MNE with consequent negative spillovers to other aliates. For
example, a bankruptcy in a group causes signicant drops in external nancing, investments
and prots of other aliates and an increase in their bankruptcy probability.
3 A Model of Internal Capital Markets
3.1 Denitions
Assume that there is an MNE with : plants/subsidiaries/aliates in dierent countries. To
nance investment, each aliated rm raises external debt from the local capital market.
In addition, the MNE can operate an internal capital market and lend to subsidiaries with
the largest need for investment funds. Alternatively, internal debt may be used to minimize
the global tax bill by shifting prots to low-tax locations. The net value of the MNE is
\
1
=

)
\
)
where \
)
refers to the value of a single, wholly-owned aliate in country ,, and
index H refers to the headquarters. The MNE is endowed with total equity/own funds

)

)
from previous operations. The distribution of own funds
)
across locations is historically
determined.
If investment 1
)
in plant , is successful, it yields an end-of-period value 1
)
+ 1
)
, where
1
)
is undepreciated capital, 1
)
= o
)
) (1
)
) is the cash-ow function satisfying )

0 )

,
and o
)
denotes productivity. Assuming that investment opportunities exceed own funds,
1
)

)
, the subsidiary has to raise internal debt 1
)
and external funds. The latter consist
of passive bank credit 1
)
and active, informed monitoring bank credit `
)
. Hence, investment
in location , is nanced by 1
)
=
)
+ 1
)
+ 1
)
+ `
)
. Total plant value is split among the
aliates stakeholders, \
)
= \
S
)
+ \
1
)
+ \
1
)
+ \
A
)
, where the upper indices refer to end-of-
period values of subsidiary dividends, internal debt, external debt from passive banks, and
external debt from monitored, active banks.
Denote the loan rates charged by standard banks by i. Since monitoring is costly and
might vary across countries, loan rates charged by active banks could dier as well, despite
of an integrated savings market with a common deposit rate. Therefore, denote the loan
1
Agency problems within rms could be of dierent nature, leading to underinvestment due to credit
rationing or overinvestment due to self-serving managers and misuse of excess internal funds (Jensen, 1986).
Lamont (1997) points out that interdependence of aliate investments may result from either type of agency
problems. Our key results do not essentially depend on which paradigm we adopt.
5
rates charged by monitoring banks in , by i
A
)
. The subsidiary is successful with probability
j and fails with probability 1 j. If it fails, cash-ow is zero and no debt, neither internal
nor external, is repaid. The deposit market yields a safe return : 0, and we utilize the
acronym 1 1+:. The loan rate on risky debt results from the no-arbitrage (or zero-prot)
condition j (1 + i) = 1. Accordingly, the end-of-period value of the subsidiary is determined
as
\
S
)
= j
[
1
)
+ 1
)
(1 + i) (1
)
+ 1
)
)
(
1 + i
A
)
)
`
)
1
)
]
1
)
. (1)
while the values \
1
)
= j (1 + i) 1
)
11
)
and \
1
)
= j (1 + i) 1
)
11
)
refer to internal debt
and external debt from passive banks, respectively, and \
A
)
= j
(
1 + i
A
)
)
`
)
1`
)
c:
)
1
)
is the value of monitored external debt.
Financial development refers to active oversight associated with informed capital. Active
banks incur monitoring costs c:
)
1
)
proportional to investment, where c is a cost parameter
and :
)
stands for the monitoring intensity. Adding up the value of all stakeholders yields
\
)
= j (1
)
+ 1
)
1
)
) (1 + c:
)
) 1
)
. (2)
With competitive lending, \
1
)
= \
1
)
= \
A
)
= 0, the subsidiary gets the entire joint surplus,
\
S
)
= \
)
, which is repatriated to the parent as an end-of-period dividend.
2
If lending
involves monitoring costs, banks must charge a higher rate, determined by j
(
i
A
)
i
)
`
)
=
c:
)
1
)
so that i
A
)
i.
The corporate tax liability owed in the source country is 1
)
. In line with common tax
practice, interest on debt is deductible while the return on equity
)
is not, so that
1
)
= t
)

[
1
)
i (1
)
+ 1
)
) i
A
)
`
)
]
. (3)
If 1
)
0, the subsidiary is a borrower and receives internal debt. If 1
)
< 0, the subsidiary
is a lender in the internal capital market. When a subsidiary in a low-tax country lends
one dollar creating a small tax liability t
|
i on interest earnings to another aliate in
a high-tax country yielding large tax savings, t

i, from interest deductions there the


global tax is reduced by (t

t
|
) i. Dierences in tax rates across countries introduce a
tax motive to use internal debt in order to shift prots to low-tax locations. This tax motive
may reinforce or oset other economic motives to use internal debt to allocate scarce capital
to those units with the most protable investment opportunities.
In the internal capital market, lending and borrowing over all subsidiaries must balance
and add up to zero. The total NPV of the MNE rm over all its operations is
\
1
=

)
\
)
:.t.

)
1
)
= 0. (4)
where \
)
= j
[
(1 t
)
) (1
)
i1
)
) (1 t
)
)
(
i
A
)
i
)
`
)
t
)
i
)
]
results upon substituting
(3) and zero-prot conditions together with 1
)
=
)
+1
)
+1
)
+`
)
. We will emphasize the
case :
)
= 0 and i
A
)
= i, implying \
)
= (1 t
)
) j (1
)
i1
)
) t
)
ij
)
, where the last term is
due to the tax disadvantage of equity.
2
There is a slight abuse of language. Strictly speaking, \

is a surplus of end-of-period wealth


over the opportunity cost 1

. The expected dividend payment to the parent rm amounts to


j
[
Y

i (1

+1

) i

]
. In addition, the parent gets expected interest payments ji1

.
6
3.2 Investment and External Borrowing
Timing: To maximize the total value in (4), the MNE makes a sequence of decisions. It
(i) allocates loans 1
)
0 on the internal capital market; (ii) raises external debt 1
)
and `
)
from local banks and sets aliate investments 1
)
; (iii) induces managerial and monitoring
eorts in all units; and (iv) pays back external funds and repatriates dividends. We solve
backwards and begin with stage (iii).
Eort: We assume that the headquarters choose eort to manage : subsidiaries in the total
conglomerate. The success probability j is high if managerial eort allocated to subsidiary
, is high. It is low (j
1
< j) if the headquarters neglect the subsidiary (low eort) and,
instead, consumes a private benet
)
1
)
proportional to investment. The parent obtains the
expected surplus of the subsidiary (income over opportunity cost of funds), consisting of the
sum of repatriated dividends and repayment of internal debt,

)
\
S
)
+ \
1
)
= j
[
1
)
+ 1
)
1
)
(1 + i) 1
)

(
1 + i
A
)
)
`
)
]
1(
)
+ 1
)
) . (5)
The subsidiarys success probability is high only when incentives are strong. The bank
must restrict external lending and not claim too high a repayment to keep the rm incen-
tivized with high enough residual earnings. The incentive constraint to assure high eort
(avoid private benets) is

)
(j)

)
(j
1
) +
)
1
)
. Substituting (5) yields
j
[
1
)
+ 1
)
1
)
(1 + i) 1
)

(
1 + i
A
)
)
`
)
]

)
1
)
.
)

)
j, (j j
1
) . (6)
The parents total stake must at least amount to
)
1
)
to be incentive compatible.
Monitoring is not contractible either. Its purpose is to reduce private benets from a
high

)
to a low
)
. Suppose (6) is satised if both monitoring and managerial eort are
high. The monitor keeps
n
)
(j) =
[
j
(
1 + i
A
)
)
1
]
`
)
. If she diverts resources c:
)
1
)
and
fails to monitor,
)
rises to a high value

)
, thereby violating the incentive constraint (6)
and discretely reducing the rms success probability to a low j
1
< j. With low eort, she
would thus get a lower expected income
n
)
(j
1
) =
[
j
1
(
1 + i
A
)
)
1
]
`
)
but obtain income
c:
)
1
)
from diverted resources. To assure monitoring, the contract must satisfy the incentive
constraint
n
)
(j)
n
)
(j
1
) + c:
)
1
)
, or
j
(
1 + i
A
)
)
`
)
j
)
1
)
. j
)
c:
)
j, (j j
1
) . (7)
External Financing and Investment: Investment and borrowing in stage (ii) are gov-
erned by optimal contracts. Formally, banks compete with contracts that maximize the rms
surplus subject to incentive constraints for managerial and monitoring eort and participa-
tion (zero prot) constraints for active and passive banks. In the optimum, all constraints
are binding. Hence, (7) and the active banks participation constraint yield
`
)
=
j
)
c:
)
1
1
)
=
j
1
j j
1
c:
)
1
1
)
. 1 + i
A
)
= (1 + i)
j
)
j
)
c:
)
. (8)
7
In equilibrium, the rm thus raises a xed fraction of investment as monitoring capital for
which it pays i
A
)
i. Any need for further external funding must come from residual bank
credit 1
)
= 1
)

)
1
)
`
)
. Substituting this and the monitors zero-prot condition into
the incentive constraint (6) shows how the rm can raise value by expanding investment.
The rm optimally invests until the constraint becomes binding,
\
)
= j [(1 t
)
) (1
)
i1
)
) t
)
i
)
] (1 t
)
) c:
)
1
)
=
)
1
)
1(
)
+ 1
)
) . (9)
Given a level of own funds
)
of the subsidiary and a level of internal debt 1
)
allocated
by the MNE, scaling up investment is possible only if the rm is able to raise additional
external debt, 1
)
. When more debt-nanced investment raises the r.h.s. of (9) at a faster
rate than the l.h.s., see Assumption (A) below, the constraint eventually becomes binding.
At that level of lending, the rm has then exhausted its debt capacity. Banks cannot extend
more credit since this would violate the incentive constraint and lead to a discrete reduction
in the success probability. If this would occur, either the bank could not break even or the
MNE would suer a discrete reduction in the joint surplus \
)
. Hence, the level of investment
and external borrowing is implicitly determined by the binding constraint (9) as a function
of internal debt, the tax rate, and other parameters. Figure 1 illustrates the solution for
the case :
)
= 0 and i = i
A
. The curved line corresponds to the l.h.s., while the upward
sloping, straight line stands for the r.h.s. of (9). If the nancing constraint does not bind,
subsidiary investment is at the rst-best level that maximizes the expected value on the
l.h.s., o
)
)

(
1
11
)
)
= i, so that the marginal return is equal to the user cost of capital i. Taxes
are not distorting the user cost since investment is nanced with 100% debt at the margin
with interest being fully deductible.
Insert Figure 1 here
If agency costs
)
are suciently large, and since high agency costs reduce the rms
pledgeable income, investment becomes constrained and is no longer determined by the user
cost formula o
)
)

(1
)
) = i. When raising more external debt to expand investment, rms hit
the nancing constraint before reaching the rst-best level. Being constrained, they earn an
excess return on capital, j
)
0. At the intersection point in Figure 1, the slopes satisfy

)
j
)
. For this to be a well-determined equilibrium, we assume
1 + j
)

)
j
)
(1 t
)
) [j (o
)
)

(1
)
) i) c:
)
] 0. (A)
Figure 1 illustrates how the parent can relax the subsidiarys nancing constraint and
expand investment in location , if it allocates more internal debt. Other determinants of
investment are plant productivity o
)
, the prot tax rate t
)
, and agency costs
)
= (|
)
. :
)
).
A rms nancing capacity depends positively on the quality of the legal and institutional
environment (variable |
)
) and on the monitoring intensity of active banks (variable :
)
).
Tighter accounting standards, for example, make management more accountable, narrow
down the possibility to shirk and enjoy private benets and are, thus, associated with lower

)
. Active oversight by informed intermediaries similarly reduces private benets and boosts
8
debt capacity. In other words, more intensive monitoring or institutional improvements
reduce private benets by

)
= (|
)
. :
)
) . d
)
,d|
)
= 1. d
)
,d:
)
= o. (10)
To obtain analytical results, we need to take the total dierential of (9),
3
d1
)
= /
)
1 d1
)
/
)
[j (1
)
i1
)
+ i
)
) c:
)
] dt
)
+ /
)
(1 t
)
) ) (1
)
) j do
)
+ /
)
1
)
d|
)
+ [o (1 t
)
) c] /
)
1
)
d:
)
.
(11)
where /
)
1, (
)
j
)
) 0. The concavity of ) (1
)
) implies that investment is concave
in internal funds, d
2
1
)
,d1
2
)
= /
2
)
1j (1 t
)
) o
)
)

)
d1
)
,d1
)
< 0. The rst inequality of
assumption (A) implies /
)
1 1. Receiving a unit of debt from other aliates boosts
investment more than proportionately, i.e., internal debt is leveraged by additional external
debt, d1
)
,d1
)
= d1
)
,d1
)
1 = /
)
1 1 0.
A higher local tax rate reduces subsidiary investment by eroding cash-ow and tightening
the nancing constraint, d1
)
,dt
)
< 0. In Figure 1, the prot curve would shift down,
leading to a reduction of aliate investment. When the subsidiary becomes more productive,
it generates higher earnings at each level of investment which boosts pledgeable income,
improves access to external nancing and expands investment, d1
)
,do
)
0. In Figure 1,
the prot curve shifts up. Tighter accounting standards, associated with more corporate
transparency and lower agency costs, boost investment, d1
)
,d|
)
0. Finally, a more active
style of business nance, captured by an exogenous increase in monitoring intensity :
)
, has
a positive and a negative eect. It raises pledgeable income and, thereby boosts investment
in proportion to o. It also imposes extra cost and makes monitored nance more expensive
which reduces investment in proportion to c. For the net eect to be positive, we must
assume o (1 t
)
) c. Since monitored nance is costly and more expensive, it might not be
demanded. Firms attract monitored nance only if it adds value. From (9), the introduction
of monitored nancing yields
d\
)
= j
)
d1
)
(1 t
)
) c1
)
d:
)
= [oj
)
(1 t
)
) c
)
] /
)
1
)
d:
)
. (12)
We must thus assume o (1 t
)
) c
)
,j
)
(1 t
)
) c for monitoring capital to be in demand.
The last inequality is implied by (A).
3.3 Internal Capital Market
Value of Subsidiary Plants: We now show how the MNE, in stage (i), allocates funds
on the internal capital market. Capital is moved to where it generates the highest return
and adds the greatest value. Figure 1 illustrates how the subsidiary value depends on the
level of internal capital received from other units. Allocating more internal debt relaxes
the nancing constraint and boosts investment. The larger scale boosts subsidiary value in
3
All derivatives for comparative static analysis are taken at an initial position of :

= 0 so that i

= i
is xed by j (1 +i) = 1 and `

= 0. We later consider the implications of nancial development, i.e., an


increase in :

, starting from zero.


9
proportion to the excess return j
)
of a constrained rm. Dierentiating (9) and using (11)
yields
\
)
1

d\
)
d1
)
= j
)
d1
)
d1
)
= j
)
/
)
1 0. (13)
\
)
11

d
2
\
)
d1
2
)
= (1 + j
)
/
)
) /
)
1 j (1 t
)
) o
)
)

)
d1
)
d1
)
< 0.
The derivative \
)
1
represents the return on internal debt, showing by how much the value of
subsidiary , rises if it receives more funds. The second equation shows that the subsidiary
value is concave in the allocated level of internal debt.
Allocating Internal Debt: We assume that subsidiaries are historically endowed with
an amount of equity or internally accumulated funds
)
. Aliate value is a function of total
capital provided internally. Using ` to denote the Lagrange-multiplier, the MNEs global
optimization problem \
1
= max
1

)
[\ (1
)
; . . .) + ` 1
)
] leads to
\
)
1
(1
)
; t
)
. o
)
. |
)
. :
)
) = ` = \
i
1
(1
i
; t
i
. o
i
. |
)
. :
)
) .

)
1
)
= 0. (14)
The MNE operates an internal capital market to allocate funds to those units where the
return is highest. The internal capital allocation is optimal when returns are equalized. As
noted in (13), the marginal value functions are downward sloping in the level of internal
debt. Figure 2 illustrates for the case of an MNE with two aliates and shows how the
allocation changes with a country-specic shock (see the next section).
Insert Figure 2 here
Condition (14) includes tax and economic motives to use internal debt. Holding every-
thing else constant, loading subsidiary i with more internal debt reduces the tax liability
and raises its value in (2) by d\
i
= jt
i
i d1
i
. If taxes are high in location i and low in ,,
i.e., t
i
t
)
, internal lending from aliate , to i (d1
i
= d1
)
) boosts MNE value by an
amount equal to global tax savings, d\
1
= j (t
i
t
)
) i d1
i
. However, internal debt also
serves economic functions and raises subsidiary value in location i by relaxing the nancing
constraint, see Figure 1. The need to do so depends on other fundamental country and plant
characteristics which may reinforce or oset the tax motive.
4 Determinants of Internal Debt
We start from a symmetric situation
)
and all other parameters are the identical. Initially,
there is no reason, neither tax nor economic, to use internal debt, i.e., 1
)
= 0. The compar-
ative static analysis reveals how certain structural changes make subsidiary establishments
dierent from local rms. Intuition suggests that those subsidiaries which face the tightest
nancing constraints (due to high taxes t
)
, inecient capital markets with little monitoring
10
:
)
, bad legal environment reected in low |
)
, and high factor productivity o
)
creating large
investment opportunities etc.), have the highest excess return and should attract the largest
internal credit. In the following analysis, we start in a situation of :
)
= `
)
= 0, leaving
j
)
= (1 t
)
) j
(
o
)
)

)
i
)
, and will often assume that nancial constraints are not too tight
and excess returns are small.
Corporate Tax: Suppose that country 1 raises the tax rate and becomes a high-tax lo-
cation. Standard reasoning suggests that local subsidiaries should attract internal debt to
save taxes. In addition, the higher tax reduces rms pledgeable income and makes them
more constrained relative to plants in other regions which creates yet another reason to shift
internal funds towards the high-tax country. Altogether, a higher tax rate should raise the
return to internal debt in that country. The derivative of (13) yields
\
)
1t

d
2
\
)
d1
)
dt
)
= (1 + j
)
/
)
) /
)
1
[
j
(
o
)
)

)
i
)
+ (1 t
)
) jo
)
)

)
d1
)
dt
)
]
0. (15)
The square bracket reects the change in the excess return, dj
)
,dt
)
. The tax directly reduces
the return, giving rise to a negative rst term. The second term is positive since lower
investment yields a higher excess return. If the nancing constraint is not too tight and the
excess return is small at the outset, the rst term is close to zero, leaving an overall positive
eect. The return on internal debt rises when subsidiary , gets taxed more heavily, \
)
1t
0.
In Figure 2, the schedule \
1
1
shifts up.
Since the higher tax diminishes the subsidiarys external nancing capacity, the MNE
gets more constrained in location 1. Given a higher return on internal debt, the MNE
makes plant 1 borrow internally (d1
1
0) from other subsidiaries which become lenders,
d1
)
< 0. Analytically, the dierential of (14) when only country 1 raises the tax rate, gives
\
1
11
d1
1
+\
1
1t
dt
1
= d` and \
)
11
d1
)
= d`. Summing over all plants and using

)
1
)
= 0
together with symmetry, \
)
11
= \
11
, yields
d`
dt
1
=
\
1
1t
:
0.
d1
)
dt
1
=
\
1
1t
:\
11
< 0.
d1
1
dt
1
= (: 1)
\
1
1t
:\
11
0. (16)
Starting from symmetry, the higher tax makes the plant in country 1 more constrained so
that the excess return becomes larger than at other locations. Internal borrowing makes the
MNE more constrained in other locations as well so that the marginal value of a unit of debt
rises by a common factor d\
)
1
= d`. This is illustrated in Figure 2.
Factor Productivity: Turning to economic fundamentals, we argue that MNEs shift
capital towards more productive plants. Higher productivity aects the return to internal
debt only via its impact on the excess return which is proportional to d
(
o
)
)

)
)
,do
)
= )

)
+
o
)
)

)
o1

o0

. The productivity shock directly boosts the excess return but the induced investment
brings it down again. If the technology is not too concave ()

small),
4
the second term is
4
Assume ) (1) = 1

, so that )

),)

=
1

. Substitute d1,d0 = / (1 t) )j and get d (0)

) ,d0 =
[
1 0)

/ (1 t) j
]
)

. With 0)

i and the multiplier / not much larger than one (consistent with a
realistic leverage factor d1,d1 = /1), the square bracket is surely positive for values of c 1,2.
11
small which establishes the positive sign in (17). When investment in location , gets more
protable compared to elsewhere, the return to internal debt rises:
\
)
10
= (1 + j
)
/
)
) /
)
1 j (1 t
)
)
[
)

(1
)
) + o
)
)

(1
)
)
d1
)
do
)
]
0. (17)
d1
1
do
1
= (: 1)
\
1
10
:\
11
0.
d1
)
do
1
=
\
1
10
:\
11
< 0.
The same steps as in (16) show that MNEs allocate funds towards more productive units.
Starting from symmetry, a higher productivity boosts the return on capital and makes the
rm more constrained. A more productive plant has a larger excess return than other ones
and, therefore, oers a higher return on internal funds. For this reason, the MNE makes
plant 1 borrow internally from other, less productive units. The borrowing makes the MNE
more constrained in other locations so that the marginal value of a unit of debt rises by a
common factor d\
)
1
= d`. Figure 2 illustrates, with t
1
replaced by o
1
.
The literature in public economics predominantly uses reduced-form models of internal
debt, predicting that internal lending should ow only from low to high-tax jurisdictions.
In our model, the internal lending pattern is much more complex. Consider a situation
where location 1 faces a low tax rate and lends to other units. Starting from a symmetric
equilibrium as in the preceeding subsection, this situation is created by a negative tax shock
dt
1
< 0. All other units face relatively higher tax rates and take on internal debt. Suppose
now that plant 1 becomes more productive, do
1
0. Given capital market frictions, it
cannot raise enough funds on the external capital market and needs to borrow internally to
accommodate the larger investment opportunities. Obviously, if the productivity dierence
to other units is large enough, plant 1 becomes a net internal borrower despite of it being
located in a low-tax country! One can easily compute an exchange rate (reecting the
respective trade-o), giving the size of the compensating productivity dierential required
to oset the tax rate dierential such that the level of internal debt us unchanged. Using
(16) and (17) yields
do
1
dt
1

o1
1
=0
=
1
1
,t
1
1
1
,o
1
=
\
1
1t
\
1
10
< 0. (18)
The upshot is that the pattern of internal debt not only depends on tax rate dierentials
but also on other rm and country-specic characteristics so that the ow of internal debt
cannot be predicted by looking at tax rate dierentials alone.
Institutional Development: A countrys institutional environment may be an important
determinant of capital market frictions. MNEs might then use internal debt to oset the
negative inuence. For example, better accounting standards and corporate governance
rules make management more accountable to outside stakeholders and reduce the moral
hazard problem. This improves access to the local, external capital market and reduces the
reliance on scarce internal funds. Better institutions should thus lead the MNE to release
internal resources to other aliates operating in countries with a less developed legal system.
5
5
The excess return should be high in a country with bad institutions. Antras and Caballero (2009) show
that capital does not ow from rich to poor countries despite of the return on investment there being high.
12
Institutional improvement reduces the return on internal debt. More formally, \
)
1|
= j
)
/
2
)
1+
(1 + j
)
/
)
) /
)
1
oj

o1

o1

o|

. With good governance, managerial incentive problems are less severe,


d
)
= d|
)
in (10), and pledgeable income is higher. In allowing for more externally nanced
investment, a better legal system would actually raise the return on internal debt, because
the plant can earn the excess return on a larger investment scale, see the term j
)
/
2
)
1.
However, this eect is small if the excess return j
)
is small. More importantly, the additional
investment erodes the return on capital and, thereby, the return on internal debt. This
negative eect is non-negligible (letting j
)
0 leaves \
)
1|
= /
)
1
oj

o1

o1

o|

< 0). On net,


\
)
1|

d
2
\
)
d1
)
d|
)
= j
)
/
2
)
1 + (1 + j
)
/
)
) /
)
1 j (1 t
)
) o
)
)

(1
)
)
d1
)
d|
)
< 0. (19)
d1
1
d|
1
= (: 1)
\
1
1|
:\
11
< 0.
d1
)
d|
1
=
\
1
1|
:\
11
0.
By improving access to local nance, institutional development reduces the return on internal
debt, i.e., shifts down the return schedule \
1
1
in Figure 2, and triggers a ow of internal debt
towards other countries with less developed institutions.
Financial Development: We study nancial development in the sense that local interme-
diaries engage in a more active, but also more costly style of lending. Intensifying monitoring
and oversight raises a rms pledgeable income and nancing capacity by reducing private
benets, d
)
= o d:
)
, see (10). On the negative side, monitoring raises intermediation
costs c:
)
1
)
which must be compensated by higher loan rates. Starting from :
)
= 0, the
introduction of monitoring capital boosts investment by d1
)
= (o (1 t
)
) c) /
)
1
)
d:
)
0
as in (11), and raises aliate value by (12). A more active local banking sector reduces the
incentive in (13) to use internal debt by
\
)
1n

d
2
\
)
d1
)
d:
)
= oj/
2
1 + (1 + j/) /1
dj
)
d:
)
< 0. (20)
The rst term is positive and arises because monitoring directly reduces agency costs which
boosts the investment multiplier. The second, negative term results because monitoring
reduces the excess return, dj
)
,d:
)
= (1 t
)
) [c + jo)

d1
)
,d:
)
] < 0, directly by raising
monitoring costs, and indirectly by boosting investment which reduces the gross return )

.
If credit constraints are not too tight, the total eect is negative. Letting j 0 leaves
\
)
1n
= /1 dj
)
,d:
)
< 0. Repeating the steps in (16) yields
d1
1
d:
1
= (: 1)
\
1
1n
:\
11
< 0.
d1
)
d:
1
=
\
1
1n
:\
11
0. (21)
Financial development relaxes the nancing constraint and makes it less protable to allocate
funds to an aliate with better access to external funds. Financial development in country
1 reduces the return on internal debt and makes aliate 1 an internal lender. Funds ow to
aliates in other regions, turning them into borrowers.
They do not address the role of MNEs to overcome local capital market problems.
13
Empirical Implications: Table 4 summarizes the most important results:
Table 4: Allocating Internal Debt
Shocks to country 1 d1
1
d1
)
Corporate income tax dt
1
0 (+) ()
Firm level productivity do
1
0 (+) ()
Institutional weakness d|
1
< 0 (+) ()
Financial underdevelopment d:
1
< 0 (+) ()
Internal debt is used to equate the marginal value of a dollar of internal funds across
locations when there are constraints on external debt. The ow of internal debt is thus
driven by tax and economic considerations. In particular, there is no clear-cut pattern any
more between tax rates and prot shifting via internal debt when countries not only dier
by tax but also along other economic fundamentals! Suppose that country 1 is a high-tax
country and also has a well developed nancial sector, t
1
t
)
and :
1
:
)
. Clearly, the
higher tax creates positive internal debt, d1
1
0, while nancial development (switching to
d|
1
0 in Table 4) leads to less internal debt, d1
1
< 0. It improves access to external funding
which allows MNEs to economize on scarce internal funds and reallocate them to other units
operating under more constrained conditions. Obviously, these two country dierences can
oset each other.
5 Data and Empirical Approach
In what follows, we seek to identify how internal borrowing responds to both local incentives
and global incentives created within an MNE which holds aliates in dierent countries. The
above model suggests that incentives to lend or borrow among units within MNEs arise from
dierences in productivity, corporate income taxation, institutional quality, and nancial
development across locations. In this section, we introduce the data and empirical approach
utilized to infer how local versus global aspects matter and illustrate that not only features
of the most-favorable market but also those of other locations matter for the internal capital
market.
5.1 Specication of Country-specic and Aliate-specic Funda-
mentals of the Internal Capital Market
Let us use o
i|
, t
i|
, |
i|
, and :
i|
to denote productivity, the corporate prot tax rate, insti-
tutional weakness, and the level of nancial underdevelopment, respectively, at the location
of aliate i in year t. Let us refer to any of these fundamentals by t
i|
{o
i|
. t
i|
. o
i|
. i
i|
},
where, to simplify matters, | (institutional quality) and : (nancial development) as used
in Sections 3 and 4 have been replaced by o = | max{|} and i = :max{:} measur-
ing institutional weakness and nancial underdevelopment, respectively, so that t
i|
is dened
such that a higher level thereof ceteris paribus raises the incentive towards internal debt
for aliate i in t. According to the theoretical model, aliate , then has ceteris paribus
14
an incentive to provide internal debt to aliate i whenever t
i|
t
)|
(see Table 4 for the
individual comparative static eects).
Let us describe informally how we will account for the incentives of the lending entities
with regard to fundamental t
i|
. Since aliate i may borrow from other aliates in dierent
countries within the same MNE according to Sections 3 and 4, we capture the aliate-
i-specic incentive to borrow internally in relation to fundamental t
i|
by considering the
weighted (indicated by superscript u) average level of that fundamental over all aliates ,
of the same MNE with at most as favorable an environment w.r.t. that fundamental towards
internal debt nancing, t
&
i|
.
6
This average fundamental t
&
i|
involves weights that are based on
the lending capacity each aliate of an MNE exhibits.
7
For a formal treatment, let us dene a binary variable
/
i,i)|
=
{
1. if t
i|
t
)|
,
0 otherwise.
(22)
which, in words, is unity if fundamental t is more favorable to internal debt nancing at
is location than at ,s in year t and zero otherwise. Furthermore, let us dene the set of
aliates 1
i|
, which consists of all units that belong to the same MNE as i in year t, except
for ones in the same country as i. Furthermore, let us denote the internal lending carried
out by unit , in t with )
)|
. Then, we may dene the relevant share or weight of , in the
lending capacity i has access to within the same MNE among all aliates in year t as
u
i,i)|
=
{
o
,.I
)
I

1
.I
o
,.I
)
I
. if /
i,i)|
)
)|
0,
0 otherwise.
(23)
The incentive of any aliate i of an MNE in year t to use internal debt nancing arises
(ceteris paribus) from the fundamentals, t
i|
, and the fundamentals at other locations of the
same rm, t
&
i|
, where the latter are dened as
t
&
i|
=
{
t
i|
. if t
i|
= min
)
(t
)|
),

)1
.I
u
i,i)|
t
)|
otherwise.
(24)
To acknowledge the lending capacity within each MNE comprehensively

)1
.I
runs over all
units, including the German parent, even though the parent does not surface in the aliate-
level data on internal debt below. Suppose aliate i faces the least favorable fundamental t
i|
in the group in year t for internal debt nancing. Then, by design of (24), t
&
i|
= t
i|
. Hence,
the dierential t
i|
t
&
i|
0 is a compact measure of the (ceteris paribus) incentive to use
cross-border internal debt according to t
i|
.
8
Notice that all fundamentals t
i|
and t
&
i|
vary over
time for the average aliate which permits conditioning on aliate-specic xed eects and
to identify the parameters on t
i|
and t
&
i|
from the time dimension within aliates.
6
Recall that one specic aspect of the above model is that, from a specic fundamentals point of view,
there is an incentive to use cross-border internal debt from aliates in countries with a less favorable
environment, depending on the conguration of other fundamentals. We will come back to this issue below.
7
Internal lending of all German aliates is observed in the data (see Section 5.3).
8
Of course, other fundamentals may generate incentives for internal borrowing or lending beyond a specic
i

in the theoretical model in Sections 3 and 4.


15
5.2 Capturing Incentives for Internal Debt with Several Fundamen-
tals
If countries or, generally, units within a rm diered only in one dimension say, the
corporate prot tax rates they face the internal capital market would be determined by a
bivariate model where only aliate is and the minimally taxed aliates tax rates would
matter. With several fundamentals for the internal capital market as in the theoretical
model outlined in Sections 3 and 4, we arrive at a multivariate empirical model. Moreover,
there is a trade-o between more favorable market conditions in some dimensions (prot
taxes, institutions, nancial development, or productivity) and less favorable ones in other
dimensions. Hence, the sharp rules for relevant borrowing or lending relationships for the
fundamentals t
i|
{o
i|
. t
i|
. o
i|
. i
i|
} in (22) do not have to hold at the stated precision. This
is in line with the model in Sections 3 and 4.
There are several options to relax those conditions, and we chose to implement heuristic
degrees of imprecision by modifying (22) to
/
i,i)|
=
{
1. if t
i|
t
)|
:o
i,i
,
0 otherwise.
(25)
where o
i,i
is the standard deviation of fundamental t
i|
calculated across all observations
for the MNE that aliate i belongs to, and : {0.1. 0.5. 1.0} is a heuristic scaling factor.
Hence, we relax the sharp conditions for internal debt nancing in (22) by up to one standard
deviation of the respective fundamental for each aliate i in year t. Notice that, in line with
the model in Sections 3 and 4, the approach in (25) implies that units in a rm may not only
borrow from ones with, say, a favorable (lower) tax rate but also ones with a less favorable
(higher) tax rate, as long as the disincentive to borrow entailed by the tax rate alone is
compensated by other favorable characteristics.
5.3 Data
The empirical investigation relies on the MiDi database (Microdatabase Direct Investment)
collected by Deutsche Bundesbank. Two aspects of MiDi are particularly noteworthy. First,
above a minimum reporting threshold, we observe all foreign aliates in Germany.
9
The fact
that MiDi reports the universe of German MNEs is especially important for our analysis, as
it allows determining tax and other incentives to use internal debt in a comprehensive way
across all units, taking into account MNEs activities in almost all countries of the world,
including all aliates and the German parent rm. Second, MiDi does not only provide
information about the aliates total debt but also about internal borrowing. Specically,
rms have to report liabilities to aliated enterprises linked with the party required to report
9
All German rms and households which hold 10% or more of the shares or voting rights in a foreign
enterprise with a balance-sheet total of more than 3 million Euros are required by law to report balance-
sheet information to the Deutsche Bundesbank. Indirect participating interests have to be reported whenever
foreign aliates hold 10% or more of the shares or voting rights in other foreign enterprises. The reporting
requirements are set by the Foreign Trade and Payments Regulation. For details and a documentation of
MiDi, see Lipponer (2009).
16
through participating interests (see Lipponer, 2009). We are particularly interested in the
determinants of cross-border internal borrowing and lending, which is supposedly used by
MNEs to shift prots. Hence, we focus on cross-border transactions among units (aliates
and the parent company) excluding all debt from aliated entities that are located in the
same country as the borrowing entity.
10
The data-set available to us comprises 45,608 aliates of German MNEs over the period
1996 to 2007. Altogether, our empirical analysis is based on 227,558 observations over this
time span. Since we have information on all internal debt provided to aliates by the parent,
we can determine the total amount of internal lending of the German parent.
According to the theoretical model, the aforementioned fundamental determinants of the
internal capital market should be included in an empirical model of internal debt nancing.
Our empirical analysis includes the following variables as determinants of internal debt of
aliate i: (i) corporate income tax (host) denotes the statutory income tax rate faced by
aliate i in year t, t
i|
; (ii) weighted corporate income tax (other locations) is the lending
capacity-weighted corporate income tax rate as dened in equation (24), t
&
i|
; (iii) nancial
underdevelopment (host) is a variable that captures the nancial underdevelopment at the
location of i in year t, i
i|
; (iv) weighted nancial underdevelopment (other locations) is the
lending capacity-weighted nancial underdevelopment dened akin to equation (24), i
&
i|
; (v)
institutional weakness (host) is a variable that captures the institutional weakness of the
host country of i in year t, o
i|
; (vi) weighted institutional weakness (other locations) is the
lending capacity-weighted institutional weakness at the other locations as dened in equation
(24), o
&
i|
; (vii) aliate-level productivity OP (host) is aliate is productivity as estimated
by the method of Olley and Pakes (1996), o
i|
; (viii) weighted aliate-level productivity (other
locations) is the lending capacity-weighted productivity (Olley and Pakes, 1996) from other
aliates within the MNE that aliate i belongs to in year t as dened in equation (24),
o
&
i|
.
11
Beyond the variables suggested by our theoretical model, even other factors might aect
internal debt. For some aliates, the costs of external borrowing, i.e., issuing bonds or
borrowing from banks, might be comparatively low. One explanation for the better access
to external debt may be that aliates dier in their opportunities to borrow against col-
lateral. Therefore, we include the variable Tangibility which reects the xed-to-total-asset
ratio of aliate i in year t. Asset tangibility is associated with higher liquidation values
and can facilitate debt nancing since a possible liquidation of a rm gets less costly for
10
In the context of our model, reasons for internal borrowing from other aliates in the same market
could only be productivity dierences, since other aspects (prot taxation, institutional standards, nancial
development) are similar for such units.
11
Note that in robustness tests we use alternative specications for the productivity variable. The estimator
of Olley and Pakes (1996) accounts for three salient features of technology: endogenous input choice by rms;
dynamics (sluggish adjustment of productivity); and selection (variability of total factor productivity of
stayers versus exiters with selection of inecient rms into exiting). We utilize (i) aliate-level log turnover
(value added is not reported in the data) as the dependent variable, (ii) log number of employees and log
nancial and other assets as inputs (intermediate goods are not recorded in the data), (iii) aliate age and
log xed assets as state variables, (iv) log net investment (the rst dierence of xed assets) as the proxy
variable which is monotonically related by assumption to the unobserved total factor productivity state
variable. Throughout, we condition on a time trend as an additional control variable.
17
shareholders as well as for debt holders. Higher liquidation values may also facilitate more
eective management control since a liquidation threat becomes more credible.
12
Harris and
Raviv (1990) nd a positive correlation between companies liquidation value (proxied by
the fraction of tangible assets) and the optimal debt level. A positive eect of tangibility on
leverage is also conrmed by Rajan and Zingales (1995), who investigate the determinants of
capital structure of public rms in G-7 countries. Bernardo, Cai, and Luo (2001) emphasize
another case that might explain a positive relationship between tangibility and internal debt.
They argue that clear repayment and interest rules can solve information problems associ-
ated with long-term investment projects. A high share of xed assets (high tangibility) may
indicate that the share of long-term investment projects is high and, therefore, internal debt
associated with regular interest payments may be the preferred source of nance. Tangi-
bility may as well capture another aspect that generally aects the use of debt. De Angelo
and Masulis (1980) point out that non-debt tax shields such as depreciation allowances
or investment tax credits associated with xed assets may crowd out the value of interest
deduction. Accordingly, tangibility may also negatively aect the internal-debt-to-capital
ratio of a rm since alternative opportunities to reduce the corporate tax burden apart from
debt are available.
Furthermore, we include the variable Loss carryforward, which is dened as a binary
variable that is unity if aliate i carries forward losses in period t and zero otherwise.
Since taxable prots in the current period can be credited against losses carried forward,
the benets of additional interest deductions may be crowded out as additional interest
payments only result in new losses that can be carried forward into consecutive periods
(see, e.g., MacKie-Mason, 1990). For this reason, we may expect that Loss carryforward
is negatively related to internal debt. If, however, a loss carryforward indicates that the
aliate is nancially distressed, more internal debt might be provided by the parent or
other aliated entities to support the rm (see Gopalan et al., 2007).
We also use Sales to capture the size and cash ow of aliate i in year t. With either
interpretation, higher sales are associated with more favorable lending conditions in terms
of external debt (see Graham and Harvey, 2001). In addition, higher sales may also imply
that a rm is more capable to retain earnings. Both arguments suggest a negative impact
on internal debt.
The fact that MiDi provides panel data allows us to control for aggregate common year-
specic eects. This captures not only simultaneous aggregate shocks in host countries
but also changes in German taxing and lending conditions as all parent rms are based in
Germany. Another advantage of panel data is that we can use aliate-specic xed eects
to control for all unobservable time-invariant factors of inuence on an aliates internal
debt. This might be important as dierent aliates can have dierent optimal internal debt
ratios, depending on aliate-specic unobservable costs and benets.
The rst and second moments of all dependent and independent variables used in the
regressions with the lending capacity or the internal debt ratio as dependent variables are
12
At the cost of complexity, one might include a liquidation value a1, a < 1, in the model which can
be accessed by creditors in case of failure and therefore allows additional debt nancing. Higher tangibility
would be associated with a higher a.
18
summarized in Table 1.
Include Tables 1 and 2 here
As already emphasized above, the unique features of the MiDi data allow us to identify
virtually all relevant activities of German MNEs abroad. Table 2 provides some information
about the geographical distribution of foreign aliates of German MNEs. In terms of the
total number of observations, the table shows that the USA, France, and the United Kingdom
are the most important host countries to German MNEs.
5.4 The Internal Debt Ratio as a Fractional Response Variable
An aliates internal debt ratio is necessarily bounded between zero and one. A linear re-
gression model will not generally obey those bounds and involve similar problems as linear
probability models do. This calls for an appropriate estimation technique, where the marginal
eect of any explanatory variable is not constant over the support region (see Papke and
Wooldridge, 1996, for further discussion). We follow Papke and Wooldridge (2008) in esti-
mating a panel-data fractional response model with the debt ratio as the dependent variable,
rst assuming strict exogeneity of all regressors, and then allowing lending capacity which
is used in the weights u
i,i)|
in (23) to be endogenous.
The response variable is the cross-border internal-debt-to-total-capital ratio for aliate
i at time t, denoted by 11
i|
[0. 1].
13
We assume the conditional expectation of 11
i|
to be
1[11
i|
x
i|
. c
i
] = (x

i|
+ c
i
). (26)
where () is the standard normal cumulative distribution function, x
i|
is a column vector
of explanatory variables, is the corresponding column vector of parameters to be esti-
mated, and c
i
is a time-constant aliate-specic unobserved eect which is allowed to be
correlated with all explanatory variables. x
i|
includes the fundamentals t
i|
and t
&
i|
for all
t {o. t. o
i|
. i
i|
} as introduced in Subsection 5.1. x
i|
also includes aliate-specic control
variables as introduced at the end of Subsection 5.3 and time dummy variables.
With regard to the modeling of c
i
, we follow the so-called
Mundlak-Chamberlain-Wooldridge device (see Mundlak, 1978, Chamberlain, 1982, 1984,
and Wooldridge, 2002) as in Papke and Wooldridge (2008) and assume that c
i
is normally
distributed conditional on x
i|
. They dene
c
i
= +x

i
+ c
i
. c
i
x
i
Normal(0. o
o
). (27)
where x
i
1
1

T
|=1
r
i|
is a column vector of the time-averaged explanatory variables for
aliate i and o
o
is the conditional variance of c
i
. Plugging this expression into (26), we
obtain
1[11
i|
x
i|
. c
i
] = ( +x

i|
+x

i
,(1 + o
o
)
12
) (28)
(
o
+x

i|

+x

) (29)
13
Note that the data allow us to focus on cross-border internal borrowing. Hence, 11

does not include


internal borrowing from aliates active in the same country.
19
and see that is identied up to the positive scalar (1 + o
o
)
12
(see Papke and Wooldridge
2008, p. 123, for details). Average partial eects can then be estimated by taking derivatives
of `
1

.
i=1
(

o
+x

i|

+x

) with respect to the variable of interest. In general (and with


the only exception of one sensitivity check where we account for country-year clustering of
the disturbances), we apply a panel bootstrapping procedure (see Fitzenberger, 1998; Papke
and Woolrdidge, 2008).
5.5 Allowing for Endogenous Lending Capacity Weights
Since internal lending and borrowing within MNEs are simultaneously determined, u
i,i)|
for
t {o. t. o
i|
. i
i|
} and, in turn, t
&
i|
are likely endogenous to the internal debt ratio of aliate
i at time t. To solve this problem, we project actual lending, )
)|
, on characteristics thereof
(including aliate-specic xed eects) to obtain a measure of predicted lending capacity,

)
)|
, yielding predicted weights, u
i,i)|
.
Notice that )
)|
is a non-negative variable which may be zero. Therefore, we use an
exponential regression model to estimate

)
)|
, conditional on aliate-specic and market-
specic time-variant variables collected in the vector z
)|
and an aliate-specic eect c
)
,
1()
)|
z
)|
. c
)
) = crj(z

)|
)c
)
. (30)
where is a vector of unknown parameters on z
)|
. We again follow the Mundlak-Chamberlain-
Wooldridge device to specify 1(c
)
z
)|
) = crj(z

)
), where z

)
are the aliate-level means of
the regressors and is a corresponding vector of unknown parameters. Then, we substitute
)
)|
in (23) by

)
)|
to calculate lending capacity weights and weighted fundamentals t
&
i|
for
t {o. t. o
i|
. i
i|
} as dened in (24).
Below, we will make use of t
&
i|
to instrument t
&
i|
.
14
We follow Papke and Wooldridge
(2008) and estimate a reduced form for t
&
i|
using the instrument t
&
i|
along with the exogenous
explanatory variables in (26). We then include the predicted residuals of this regression,
denoted as
i|
, to control for the potential endogeneity of t
&
i|
in the fractional response model
(26).
15
6 Results
6.1 Baseline Results for Exogenous versus Endogenous Lending Ca-
pacity
Table 3 shows the baseline results for a fractional response model on the internal debt ratio
11
i|
as in (26), assuming that all (lending capacity-)weighted fundamentals are strictly ex-
ogenous. The fractional response model is estimated by pooled quasi maximum-likelihood
14
Note that i

carries the information of the exogenous variables z

which capture conditions at the


lending aliates.
15
See Papke and Wooldridge (2008, p. 125) for a detailed discussion of the procedure.
20
estimation (QMLE).
16
Note that all regressions include time- and aliate-specic xed ef-
fects.
Include Table 3 here
The rst column of the table conrms results of earlier studies, suggesting that the local
statutory tax rate positively relates to the share of internal cross-border debt of aliate
i. The columns labeled Coe. contain the estimated coecients (and standard errors in
parentheses), while the columns labeled APE display the re-scaled coecients as average
partial eects, which may be compared to the coecients of a linear model. Once we include
the weighted tax rate of the lending part of an MNE, the coecient of the local tax eect
becomes slightly smaller but remains positive and signicantly dierent from zero.
Note that the point estimates on all other incentive variables (or fundamentals) suggested
by our model have the expected sign. According to our denition of those variables, the local
incentive to internal debt nancing should increase with a higher value of the respective
variable (t
i|
). Consistent with that, the weighted foreign fundamentals (in other aliates
and countries than i; t
&
i|
) should exert a negative eect. While the estimate for the weighted
corporate income tax (other locations) is not signicantly related to internal debt, we do
not want to overemphasize the ndings of Table 3, because all variables portraying foreign
fundamentals are weighted by actual lending, and the estimates are likely biased as indicated
in Subsection 5.5.
Include Table 4 here
Table 4 allows for endogenous weighted regressors, using the approach suggested by Papke
and Wooldridge (2008) and described in Subsection 5.5. Other than that, the specication
underlying Table 4 is identical to the one in the last two columns of Table 3. One interesting
nding in comparison to the earlier results is that the magnitudes of the host coecients
(on t
i|
) and the other-location coecients (on t
&
i|
) are similar in Table 4 for all fundamentals
t. In comparison, the corporate tax eects were much more asymmetric when assuming
exogeneity in Table 3. Moreover, internal cross-border debt seems to be even much more
strongly determined by dierences in corporate taxes across locations when allowing for
endogenous eects rather than assuming exogeneity.
The average partial eect (APE) of the corporate income tax in the host country (t
i|
),
which is displayed in column 2 of Table 4, implies that a ten percentage point higher local
corporate income tax rate leads to a 9.2 percentage point higher internal-debt-to-capital
ratio. This is almost four times as large as the comparable eect in Table 3. We may
compare this estimate to the ndings of a meta-study by Heckemeyer, Feld, and Overesch
(2011), investigating 46 studies on the impact of taxes on debt-to-capital ratios. Their results
suggest a marginal tax eect on debt of about 0.3. Hence, the marginal eect identied in
the underlying census-type data-set is much larger than in other studies which mainly focus
on larger rms. A ten percentage point higher corporate income tax rate at other relevant
16
For a discussion on dierent estimation methods see Papke and Wooldridge (2008, p. 124).
21
lending locations of the same MNE leads to an almost symmetric negative eect of the same
magnitude on aliate is debt-to-capital ratio in the same year.
The regressions in Table 3 suggested that all of the fundamentals postulated by the
above theoretical model matter for an aliates internal debt ratio. With the exception
for productivity, the same conclusions apply when allowing for the endogeneity of lending
capacity in Table 4.
17
A more severe nancial underdevelopment and institutional weakness
in the host country lead to a signicantly higher internal-debt-to-capital ratio. Consistent
with the incentives described in Subsection 5.1, the weighted averages of these two measures
across other locations within the same rm have a negative eect on the share of internal
debt of a given aliate.
Two of the aliate-specic control variables included in the estimations are also signif-
icantly related to internal debt nancing. First, a higher share of tangible (xed) assets
(Tangibility) implies a higher internal-debt-to-capital ratio. Since a high tangibility is a
proxy for the importance of long-term investment projects in a rm, this nding might re-
ect that monitoring problems associated with investment projects are partly solved by using
internal debt (see Bernardo, Cai, and Luo, 2001). The positive coecient of the Loss car-
ryforward indicator variable conrms that internal debt is a exible source of nance that
can be provided by aliated entities (see Gopalan, Nanda, and Seru, 2007). Beyond the
mentioned variables, rm size in terms of foreign aliate sales does not enter signicantly
as a driver of internal debt.
6.2 Sensitivity Analysis with Regard to the Incentives to Use In-
ternal Debt
The results in Tables 3 and 4 do not allow for a trade-o between the dierent fundamentals
determining the incentives to use cross-border internal debt. There, the (ceteris paribus)
incentive to use internal debt according to fundamental t
i|
{o
i|
. t
i|
. o
i|
. i
i|
} is given by the
dierential t
i|
t
&
i|
0, where t
&
i|
is the weighted-average over all aliates , with t
i|
t
)|
.
This ignores that there is a trade-o between more favorable conditions in some fundamentals
(prot taxes, institutional weakness, nancial underdevelopment, or productivity) and less
favorable ones in others.
Tables 5 to 7 present results using weighted foreign fundamentals constructed as described
in (25). Here we allow the incentives in each dimension to become negative by building t
&
i|
over a larger set of aliates , with t
i|
t
)|
:o
i,i
, where o
i,i
is the standard deviation of
fundamentals t
i|
calculated across all observations within the MNE aliate i belongs to, and
: takes on the values 0.1, 0.5, and 1.0 in Tables 5, 6, and 7, respectively. That is, we relax
the sharp conditions for internal debt nancing in (22) alternatively by one tenth of, one
half of, and one standard deviation of the respective fundamental for each aliate i in year
t.
17
However, aliate-level productivity at aliate i, and even more so at (weighted) other locations within
the same rm, displays a low degree of variation over short periods of time. Hence, it is hard to discern the
productivity variables eects on the debt ratio from the one of aliate-specic time-invariant eects on the
one hand and from common time eects on the other hand.
22
The magnitude of the eects across Tables 5 to 7 is very similar to those in Table 4 and
the levels of signicance are somewhat higher. The APEs of the corporate income tax and
institutional weakness in the host country are respectively smaller and higher than in Table
4. For example, the APEs reported in Table 7 imply that a ten percentage point higher local
corporate income tax rate leads to a 7.2 point higher internal-debt-to-capital ratio, while an
increase in the index of institutional weakness of one standard deviation (1.7) increases the
internal-debt-to-capital ratio by 3.5 percentage points.
6.3 Sensitivity Analysis with Regard to Aliate-Productivity Mea-
surement
While all results presented in Tables 3 to 7 use the method suggested by Olley and Pakes
(1996) to estimate total factor productivity of an aliate, we may investigate the sensitivity
of our ndings with respect to other productivity measures. Table 8 shows that our main
results are not aected at all when doing so.
In columns 1 and 2 we estimate aliate-level productivity by using the method of Levin-
sohn and Petrin (2003).
18
For both estimated coecients (for the host-country productivity
and for the weighted productivity at other locations), we cannot nd a signicant impact
on the internal debt ratio of aliates. In columns 3 and 4 we use an alternative, less accu-
rate measure for productivity calculated as aliate sales divided by the average change in
the total assets of an aliate.
19
Again, we cannot conrm that this alternative measure is
signicantly related to the internal debt ratio of the foreign aliates.
6.4 Further Sensitivity Analysis
We have conducted further sensitivity checks regarding (i) the treatment of location choice
of aliates as to be endogenous, (ii) an alternative treatment of standard errors by using
country-year clustering instead of a panel bootstrap procedure, (iii) eects on small versus
large rms, (iv) the relative importance of conditioning on other fundamentals than corporate
prot tax rates, (v) the role of estimating nonlinear models (which respect to the bound-
edness of the dependent variable) relative to linear (OLS) models, (vi) the consideration
of interaction terms of the fundamental variables so that their nonlinear eects on internal
borrowing do not only accrue to the nonlinear functional form of the estimator for fractional
dependent variables, and (vii) the consistency of internal borrowing within a country with
rm-level productivity dierences across all units of a rm in that country and year.
18
In contrast to Olley and Pakes (1996), Levinsohn and Petrin (2003) suggest using intermediate inputs
(materials) as a proxy variable instead of investment. As said above, materials are not recorded in the
aliate statistics at hand. Therefore, we use nancial and other assets which is also more stable than net
investment as a proxy variable which is is monotonically related by assumption to the unobserved total
factor productivity state variable. In other regards, the adopted estimation procedure is similar to the one
of Olley and Pakes (1996).
19
In contrast to the estimates based on Olley and Pakes (1996) and Levinsohn and Petrin (2003), this
proxy variable does not account for the endogeneity of some of the inputs, nor does it account for selectivity.
23
A detailed presentation of the associated results is provided in a separate online appendix
for the sake of brevity of the manuscript. The main conclusions drawn from this analysis
are the following. First, a control function approach to control for endogenous aliate
location choice does not lead to qualitatively or quantitatively dierent results from the
ones presented above. Country-year clustering of the standard errors does not lead to a big
dierence in standard errors relative to the panel bootstrap. The inclusion of small rms, the
treatment of endogenous fundamentals of other aliates than i for is internal borrowing,
and the treatment of the internal borrowing (share) as a bounded dependent variable are
major reasons for the identication of much larger tax semi-elasticities in this paper than in
previous work. The consideration of interaction eects of fundamentals beyond the nonlinear
functional form of the econometric model does not add much in terms of nonlinearity of the
eects of fundamentals on internal borrowing in the support region of the data (this is aligned
with our expectations; see Greene, 2010). Finally, within-country productivity dierences
are of a similar importance for internal borrowing as productivity dierences across countries.
6.5 Quantication of Tax Eects and Their Discussion in the Light
of the Literature
The novel aspect of our investigation is that we allow MNEs to use their internal capital
markets to reallocate capital to entities facing constraints in general, and high taxes, weak
institutions, an underdeveloped nancial market, and a high productivity in host-countries in
particular. An important advantage of our study over existing empirical work is that the data
allow us to observe lending and borrowing among aliates of MNEs in a comprehensive way,
because German MNEs are required to report their capital links to Deutsche Bundesbank
so that we may gain a virtually complete picture of the internal capital market of a group.
Since the literature has devoted much attention on tax incentives for internal debt, let us
compare our ndings to previous estimates for the impact of taxes on debt. Feld, Heckemeyer,
and Overesch (2011) identify in a meta-study, synthesizing evidence from 46 studies, a typical
semi-elasticity of 0.313 of total debt. Our estimated host-country tax coecient of .918 (see
Table 4) translates into a semi-elasticity of 5.02, which exceeds the typical semi-elasticity
found in the meta-study by a factor of more than 16. If only internal debt is considered,
the meta-analysis nds a typical semi-elasticity of 0.47, which is still less than a tenth of our
estimate. As indicated in Subsection 6.4, there are three likely reasons for the big dierence
between the estimated semi-elasticity in this paper and the ones in earlier work. First, while
other data-sets often include large MNEs only, we consider also relatively small MNEs with
rather modest internal debt ratios in our Census of data. Second, while earlier work often
used linear models we resort to a nonlinear framework which pays attention to the limited
dependent variable nature of internal-debt-to-capital ratios employed as the dependent vari-
able. Third, unlike earlier work we consider a more complete array of incentives to internal
borrowing, where multiple units in a rm may act as lenders and borrowers simultaneously.
Against the background of several papers speculating about why the tax-sensitivity of debt
is so low (see Ruf, 2011), our results suggest that internal debt is highly tax-responsive and
that the tax rate is an important determinant of the internal-debt-to-capital ratio.
24
The non-linear fractional response model used in our analysis allows us to evaluate
marginal eects at dierent values of the explanatory variables. Hence, MNEs with cer-
tain characteristics may respond even more to variations in taxes than others. Figures 4 and
5 emphasize this aspect by showing how predicted internal debt ratios vary according to vari-
ations in host-country characteristics. Tables 9 and 10 present the respective estimates for
some countries that exhibit extreme values of these characteristics in view of the distribution
of these variables across countries. Comparing Greece with the United States, for example,
shows that we would predict about the same internal debt ratio, even though the statutory
tax rate in the United States is about 10 percentage points higher. Another interesting
comparison between Greece and the United States shows that the institutional environment
faced by foreign aliates in Greece ought to be improved by 0.36 (to 6.3) to predict ceteris
paribus the same internal debt ratio for the average aliate located in Greece or the United
States. With respect to capital market development, Japan, for example, would have to
improve its capital market development index to a value of 32 (from about 125) for aliates
located in Japan to exhibit the same predicted internal debt ratio as aliates located in the
United States. A comparison between aliates in Hong Kong and aliates in Singapore,
on the other hand, shows that capital market conditions may become 2.5 (from 168 to 434)
times worse in Hong Kong until the lower tax rate in Hong Kong is oset and the aliates
rely on the same amount of internal debt nancing as in Singapore. The non-linearity of the
relationship between tax and institutional components in determining 11
i|
is also reected
in dierent gradients with respect to fundamentals, which depend on where a country is
located in tax-institution space. Beside Figures 4 (in case of the institutional environment)
and 5 (in case of capital market development), the last columns of Tables 9 and 10 demon-
strate that the eect of a marginal increase in the host-country tax rate diers a lot across
locations.
Include Tables 9 and 10, as well as Figures 4 and 5 here
Finally, suppose that we observe an MNE that consists of just two entities. The bor-
rowing entity is located in a low-tax country and the lending entity is located in a high-tax
country. At the same time, the borrowing entity is located in a country with bad institu-
tions and an underdeveloped capital market, while the lending entity is located in a country
with sound institutions and a well developed capital market. Let us assume values for the
respective fundamentals of the borrowing (lending) aliate that refer to the 95
|
percentile
(5
|
percentile) of the respective distribution of o and i, and a corporate prot tax rate at
the lending location of 40%. We may compare the predicted internal debt ratio of this MNE
to the average MNE in our sample and determine the (negative) tax rate dierential for this
example which the rm would be willing to accept to end up with the same predicted share
of internal debt. For the average MNE, our model predicts an internal-debt-to- capital ratio
of 0.18. Note that from a pure taxation point of view, aliates in low-tax countries would
not borrow from aliates in high-tax countries. In our example, however, given other fun-
damentals of the internal capital market, the borrowing aliate would be willing to accept
a negative tax rate dierential of 11 percentage point (i.e., a host-country tax rate of 29%)
and still choose an internal debt ratio of 0.18. Figure 6 describes how the tax incentives
interact in such an example.
25
Include Figure 6 here
7 Conclusions
This paper has shown how multinational rms allocate internal nancing by means of an
internal capital market not only to shift prots but also to avoid constraints faced by aliates
in foreign countries. Similar to a high productivity of these aliates, constraints associated
with a weak institutional environment or an underdeveloped capital market lead to an excess
return on investment. In this sense, the internal capital market is used to allocate nancing
to aliates with investment opportunities that entail the highest return.
Earlier work has mainly focused on tax responses of internal borrowing from the unit
where the incentive is maximal. We illustrate that internal capital markets render predic-
tions concerning tax eects more complex than portrayed before. In particular, dierences in
incentives given by fundamentals such as institutional weakness, nancial underdevelopment
or productivity can oset tax incentives so that aliates in high-tax countries may lend to
aliates in low-tax countries. For the empirical analysis we use the Microdatabase Direct
Investment (MiDi) provided by Deutsche Bundesbank, which is a unique data-set of Ger-
man multinational rms and their foreign aliates. Since German law requires mandatory
reporting to Deutsche Bundesbank, MiDi includes the universe of German multinational
rms. This feature of the data allows us to capture tax, institutional, capital market, and
productivity incentives in a comprehensive way, since not only the borrowing parties within
a group but also the lending parties are observed.
Using a fractional response model, our main results imply that a one percentage point
higher statutory tax rate in the host country is associated with a 0.92 percentage point
higher internal-debt-to-capital ratio of the borrowing aliate. At the same time, a one
percentage point increase in the (weighted) tax rate of the lending aliates is associated
with a 0.77 percentage point lower internal-debt-to-capital ratio of the borrowing aliate.
Other determinants of internal debt show the same qualitative pattern. While nancial
underdevelopment in the host country is positively related to the internal debt ratio of the
borrowing aliate, nancial underdevelopment at other locations exerts a negative eect on
the internal debt ratio of the borrowing aliate. While nancial weakness in the host country
is associated with a higher internal debt ratio of the borrowing aliate, nancial weakness
at the lender locations leads to a lower internal debt ratio of the borrowing aliate. While
a higher aliate-level productivity of the borrowing aliate relates positively to its internal
debt ratio, a higher productivity of the lenders exerts a negative impact on the internal debt
ratio of the borrowing aliate (it should be mentioned, however, that productivity measures
are found to be statistically insignicant).
Since this paper shows that internal debt within multinational rms is not only used
to avoid taxes but also to compensate dierences in other fundamentals, tax policy must
consider that anti-tax avoidance measures designed to restrict prot shifting of multinational
rms (e.g. thin-capitalization rules) might aggravate nancing constraints caused by non-
tax fundamentals. Given our ndings, such policies would have signicant eects on real
investment decisions of multinational rms, which go beyond their actual purpose.
26
Acknowledgements
We are particularly grateful to Giorgia Mani and Joel Slemrod for their discussions and
numerous helpful comments on an earlier version the paper. Moreover, we gratefully acknowl-
edge comments by Rita de la Feria, Michael Devereux, Michelle Hanlon, Jim Hines, Nicholas
Sly, and Alfons Weichenrieder. Finally, we acknowledge comments of other participants at
the annual meeting of the CESifo Workshop on Corporate Taxes and Corporate Governance
at Munich, the European Economic Association at Mlaga, the Finanzwissenschaftlicher
Ausschuss of the German Economic Association at Vienna, the NBER TAPES Conference
at Oxford University, and at a research seminar University of Tbingen.
27
References
[1] Antras, Pol, Mihir A. Desai and Fritz C. Foley (2009), Multinational Firms, FDI Flows,
and Imperfect Capital Markets, Quarterly Journal of Economics 124, 1171-1219.
[2] Altshuler, Rosanne and Harry Grubert (2003), Repatriation Taxes, Repatriation Strate-
gies and Multinational Financial Policy, Journal of Public Economics 87, 73107.
[3] Alworth, Julian and Giampaolo Arachi (2001), The Eect of Taxes on Corporate Fi-
nancing Decisions: Evidence from a Panel of Italian Firms, International Tax and Public
Finance 8, 353367.
[4] Bolton, Patrick and David S. Scharfstein (1998), Corporate Finance, the Theory of the
Firm, and Organizations, Journal of Economic Perspectives 12, 95-114.
[5] Booth, Laurence, Varouj Aivazian, Asli Demirguc-Kunt, and Vojislav Maksimovic
(2001), Capital Structures in Developing Countries, The Journal of Finance 56, 87
130.
[6] Buettner, Thiess, Michael Overesch, Ulrisch Schreiber and Georg Wamser (2009), Tax-
ation and Capital Structure Choice - Evidence from a Panel of German Multinationals,
Economics Letters 105, 309-311.
[7] Chamberlain, Gary (1984), Panel data, in Griliches, Zvi and Michael D. Intriligator
(Eds), Handbook of Econometrics, vol.2 North Holland, Amsterdam, 12471318.
[8] Chamberlain, Gary (1982), Multivariate Regression Models for Panel Data, Journal of
Econometrics 18, 546.
[9] Collins, Julie H. and Douglas A. Shackelford (1997), Global Organizations and Taxes:
An Analysis of the Dividend, Interest, Royalty, and Management Fee Payments between
U.S. Multinationals Foreign Aliates, Journal of Accounting and Economics 24, 151
173.
[10] Cook, Douglas O., Robert Kieschnick, and B.D. McCullough (2008), Regression Anal-
ysis of Proportions in Finance with Self Selection, Journal of Empirical Finance 15,
860867.
[11] De Angelo, Harry and Ronald W. Masulis (1980), Optimal Capital Structure under
Corporate and Personal Taxation, Journal of Financial Economics, 329.
[12] Desai, Mihir A., C. Fritz Foley and James R. Hines Jr. (2004), A Multinational Per-
spective on Capital Structure Choice and Internal Capital Markets, Journal of Finance
59, 2451-2487.
[13] Devereux, Michael P. (2007), The Impact of Taxation on the Location of Capital, Firms
and Prot: A Survey of Empirical Evidence, OUCBT WP 07/02.
28
[14] Egger, Peter, Wolfgang Eggert, Christian Keuschnigg and Hannes Winner (2010), Cor-
porate Taxation, Debt Financing and Foreign Plant Ownership, European Economic
Review 54, 96-107.
[15] Feld, Lars P., Jost H. Heckemeyer, and Michael Overesch (2011), Capital Structure
Choice and Company Taxation: A Meta-Study, CesIfo Working Paper No. 3400.
[16] Fitzenberger, Bernd (1998), The Moving Blocks Bootstrap and Robust Inference in
Linear Least Squares and Quantile Regressions, Journal of Econometrics 82, 235287.
[17] Frankel, Jerey A. and David Romer (1999), Does Trade Cause Growth?, American
Economic Review 89, 279299.
[18] Fuest, Clemens and Thomas Hemmelgarn (2005), Corporate Tax Policy, Foreign Firm
Ownership, and Thin Capitalization, Regional Science and Urban Economics 35, 508-
526.
[19] Gertner, Robert H., David S. Scharfstein, and Jeremy C. Stein (1994), Internal Versus
External Capital Markets, Quarterly Journal of Economics 109, 1211-1230.
[20] Gordon, Roger H. and Young Lee (2003), Do Taxes aect Corporate Debt Policy?
Evidence from U.S. Corporate Tax Return Data, Journal of Public Economics 82, 195
224.
[21] Gopalan, Radhakrishnan, Vikram Nanda and Amit Seru (2007), Aliated Firms and
Financial Support: Evidence from Indian Business Groups, Journal of Financial Eco-
nomics 86, 759-795.
[22] Graham, John R. (1996), Debt and the Marginal Tax Rate, Journal of Public Economics
41, 4173.
[23] Graham, John R. (1999), Do Personal Taxes Aect Corporate Financing Decisions?,
Journal of Public Economics 73, 147185.
[24] Graham, John R. (2003), Taxes and Corporate Finance: A Review, Review of Financial
Studies 16, 10751129.
[25] Greene, William H. (2010), Testing Hypotheses about Interaction Terms in Nonlinear
Models, Economics Letters 107, 291-296.
[26] Holmstrom, Bengt and Jean Tirole (1997), Financial Intermediation, Loanable Funds,
and the Real Sector, Quarterly Journal of Economics 62, 663-691.
[27] Huizinga, Harry, Luc Laeven, and Gatan Nicodme (2008), Capital Structure and
International Debt Shifting, Journal of Financial Economics 88, 80118.
[28] Jensen, Michael (1986), Agency Costs of Free Cash-Flow, Corporate Finance, and
Takeovers, American Economic Review 76, 323-329.
29
[29] Jog, Vijay and Jianmin Tang (2008), Tax Reforms, Debt Shifting and Tax Revenues:
Multinational Corporations in Canada, International Tax and Public Finance 8, 525.
[30] Lamont, Owen (1997), Cash Flow and Investment: Evidence from Internal Capital
Markets, Journal of Finance 52, 83-109.
[31] Levinsohn, J. and A. Petrin (2003), Estimating Production Functions Using Inputs to
Control for Unobservables. Review of Economic Studies 70, 317-342.
[32] MacKie-Mason, Jerey K. (1990), Do Taxes Aect Corporate Financing Decisions?,
Journal of Finance 45, 14711493.
[33] Mintz, Jack and Michael Smart (2004), Income Shifting, Investment, and Tax Compe-
tition: Theory and Evidence from Provincial Taxation in Canada, Journal of Public
Economics 88, 1149-1168.
[34] Mintz, Jack and Alfons J. Weichenrieder (2009), The Indirect Side of Direct Investment
Multinational Company Finance and Taxation, MIT Press, Cambridge, MA.
[35] Modigliani, Franco and Merton H. Miller (1958), The Cost of Capital, Corporation
Finance, and the Theory of Investment, American Economic Review 48, 261297.
[36] Mundlak, Yair (1978), On the Pooling of Time Series and Cross Section Data, Econo-
metrica 46, 6985.
[37] Olley, S., and A. Pakes (1996), The Dynamics of Productivity in the Telecommunications
Equipment Industry, Econometrica 64, 1263-98.
[38] Overesch, Michael (2009), The Eects of Multinationals Prot Shifting Activities on
Real Investments, National Tax Journal 62, 5-23.
[39] Papke, Leslie E. and Jerey M. Wooldridge (1996), Econometric Methods for Fractional
Response Variables with an Application to 401(k) Plan Participation Rates, Journal of
Applied Econometrics 11, 619632.
[40] Papke, Leslie E. and Jerey M. Wooldridge (2008), Panel Data Methods for Fractional
Response Variables with an Application to Test Pass Rates, Journal of Econometrics
145, 121133.
[41] Rajan, Raghuram G. and Luigi Zingales (1995), What Do we Know about Capital
Structure? Some Evidence from International Data, The Journal of Finance 50, 1421
1460.
[42] Ruf, M. (2011), Why is the Response of Multinationals Capital-Structure Choice to
Tax Incentives That Low? Some Possible Explanations, FinanzArchiv 67, 123-144.
[43] Scharfstein, David S. and Jeremy C. Stein (2000), The Dark Side of Internal Capital
Markets: Divisional Rent-Seeking and Inecient Investment, Journal of Finance 55,
2537-2564.
30
[44] Stein, Jeremy C. (1997), Internal Capital Markets and the Competition for Corporate
Resources, Journal of Finance 52, 111-133.
[45] Wooldridge, Jerey M. (2002), Econometric analysis of Cross Section and Panel Data,
MIT Press, Cambridge, MA.
31
Appendix Prediction of Lending Capacity
For the sake of the instrumental variable approach, we use yearly poisson regressions to pre-
dict the potential lending of each aliate. The predicted values of these regressions are then
used to weight the tax rates of the lending countries. The yearly regressions to predict lend-
ing include the following variables: corporate income tax rate, nancial underdevelopment,
and institutional weakness of the lenders country, the aliate-level productivity, tangibil-
ity, loss carryforward and the sales of the lending aliate. The regressions also include the
aliate-specic means of all variables. Figure 7 shows the world map of actual lending from
all countries aliates of German multinationals are located in. Notice that this approach
is similar in spirit to the one taken by Frankel and Romer (1999), who estimate a bilateral
gravity model of international trade and employ the (aggregated) predictions of that model
to instrument (unilateral) observed trade as a determinant of economic growth. In our case,
predicted lending serves as an instrument for observed lending which then is used to aggre-
gate up all non-i aliate fundamentals as weighted averages where the weights are allowed
to be endogenous.
This approach leads to a just-identied instrumental variables approach akin to Wooldridge
(2002; chapter 18) so that we can not apply over-identication tests. However, since lending
is predicted by fundamentals that can not easily be changed by an aliate from year to
year and we condition on time-specic eects through xed eects, it is plausible to assume
that the instruments are exogenous. In any case, the predicted weighted aliate funda-
mentals turn out to be jointly highly relevant determinants of observed weighted aliate
fundamentals as is indicated by the F-statistics reported in the respective table footnotes.
32
Appendix Figures
0
t u t = [(1 )( ( ) ) ]
j j j j j j j
V p f I iI iA
<
FB
j j
I I
> > 0
j j
j
D
+ ( )
j j j j
I D A R
j
V
Figure 1: Investment and Internal Debt
33

1
plant 1
plant 2
2
D
V
1
A
2
A
D
1
D
V

Figure 2: Internal Capital Market


34
1


6
4
6
4


3
9
3
3
9
3


1
0
6
0
1
0
6
0


1
9
4
1
1
9
4
1


3
8
7
4
3
8
7
4


6
7
4
6
6
7
4
6


7
4
5
3
7
4
5
3


1
1
6
0
0
1
1
6
0
0


1
4
7
8
2
1
4
7
8
2


2
9
7
8
7
2
9
7
8
7


4
2
3
1
0
4
2
3
1
0


1
6
4
8
4
4
1
6
4
8
4
4


3
1
5
4
3
7
3
1
5
4
3
7


6
2
7
1
1
8
6
2
7
1
1
8


1
.
0
e
+
0
6
1
.
0
e
+
0
6


1
.
2
e
+
0
6
1
.
2
e
+
0
6


1
.
7
e
+
0
6
1
.
7
e
+
0
6


4
.
0
e
+
0
6
4
.
0
e
+
0
6


3
.
0
e
+
0
7
3
.
0
e
+
0
7


4
.
6
e
+
0
8
0
F
i
g
u
r
e
3
:
L
e
n
d
i
n
g
b
y
c
o
u
n
t
r
i
e
s
i
n
2
0
0
5
N
o
t
e
s
:
S
u
m
o
f
l
e
n
d
i
n
g
i
n
t
h
e
y
e
a
r
2
0
0
5
o
v
e
r
a
l
l
a

l
i
a
t
e
s
i
n
a
c
o
u
n
t
r
y
.
I
n
f
o
r
m
a
t
i
o
n
o
n
l
e
n
d
i
n
g
i
s
t
a
k
e
n
f
r
o
m
M
i
D
i
.
T
h
e
t
e
n
m
o
s
t
i
m
p
o
r
t
a
n
t
l
e
n
d
e
r
c
o
u
n
t
r
i
e
s
a
r
e
(
i
n
d
e
c
r
e
a
s
i
n
g
i
m
p
o
r
t
a
n
c
e
)
:
G
e
r
m
a
n
y
,
U
S
A
,
N
e
t
h
e
r
l
a
n
d
s
,
U
K
,
L
u
x
e
m
b
o
u
r
g
,
C
a
y
m
a
n
I
s
l
a
n
d
s
,
C
a
n
a
d
a
,
F
r
a
n
c
e
,
N
e
t
h
e
r
l
a
n
d
s
A
n
t
i
l
l
e
s
,
B
e
l
g
i
u
m
.
32
F
i
g
u
r
e
3
:
L
e
n
d
i
n
g
b
y
c
o
u
n
t
r
i
e
s
i
n
2
0
0
5
N
o
t
e
s
:
S
u
m
o
f
l
e
n
d
i
n
g
i
n
t
h
e
y
e
a
r
2
0
0
5
o
v
e
r
a
l
l
a

l
i
a
t
e
s
i
n
a
c
o
u
n
t
r
y
.
I
n
f
o
r
m
a
t
i
o
n
o
n
l
e
n
d
i
n
g
i
s
t
a
k
e
n
f
r
o
m
M
i
D
i
.
T
h
e
t
e
n
m
o
s
t
i
m
p
o
r
t
a
n
t
l
e
n
d
e
r
c
o
u
n
t
r
i
e
s
a
r
e
(
i
n
d
e
c
r
e
a
s
i
n
g
i
m
p
o
r
t
a
n
c
e
)
:
G
e
r
m
a
n
y
,
U
S
A
,
N
e
t
h
e
r
l
a
n
d
s
,
U
K
,
L
u
x
e
m
b
o
u
r
g
,
C
a
y
m
a
n
I
s
l
a
n
d
s
,
C
a
n
a
d
a
,
F
r
a
n
c
e
,
N
e
t
h
e
r
l
a
n
d
s
A
n
t
i
l
l
e
s
,
B
e
l
g
i
u
m
.
35
Figure 4: Prediction in t-o-space
Notes: Blue dots denote predictions for the countries included in the estimation sample (evaluated at mean values of
the explanatory variables and the country-specic means of t and c). Surface corresponds to the predicted internal
debt ratio for varying values of t and c.
36
Figure 5: Prediction in t-i-space
Notes: Blue dots denote predictions for the countries included in the estimation sample (evaluated at mean values of
the explanatory variables and the country-specic means of t and r). Surface corresponds to the predicted internal
debt ratio for varying values of t and r.
37
Figure 6: Prediction in t-t
n
-space
Notes: Surface corresponds to the predicted internal debt ratio for varying values of t (c-|) and t (c|c|cco|ica-).
38
Appendix Tables
Table 1: Descriptive Statistics
Variable Acronym Mean (Std. Dev.)
Internal-debt ratio 11
.I
.1827 .2564
Corporate income tax (host) t
.I
.3235 .0724
Weighted corporate income tax (other locations) t
u
.I
.3062 .0743
Financial underdevelopment (host) r
.I
211.2420 53.5578
Weighted nancial underdevelopment (other locations) r
u
.I
196.4351 52.5706
Institutional weakness (host) c
.I
3.8210 1.6974
Weighted institutional weakness (other locations) c
u
.I
3.3883 1.7441
Aliate-level productivity OP (host) 0
.I
-21.5328 2.7130
Weighted aliate-level productivity OP (other locations) 0
u
.I
-22.4540 3.0385
Aliate-level productivity LevPet (host) 0
.I
.3505 5.9113
Weighted aliate-level productivity LevPet (other locations) 0
u
.I
.1675 4.0388
Aliate-level productivity Alt (host)
(u)
0
.I
.1928 9.3171
Weighted aliate-level productivity Alt (other locations)
(u)
0
u
.I
-.0554 8.7226
Tangibility .2490 .2731
Loss carryforward .3123 .4635
Sales .0619 .5536
Notes: 227,558 observations (
(u)
217,103 observations). All aliate-level variables are taken from the
MiDi database provided by the Deutsche Bundesbank. Internal-debt ratio is the cross-border-internal-
debt-to-total-capital ratio, where total capital consists of registered capital, capital reserves and prot
reserves, as well as internal and external debt. Corporate income tax (host) is the statutory tax rate
of the country hosting the borrowing aliate i. Weighted corporate income tax (other locations) is the
lending-weighted corporate income tax rate as dened in (24). The tax data is collected from databases
provided by the International Bureau of Fiscal Documentation (IBFD) and tax surveys provided by
Ernst&Young, PwC, and KPMG. Financial underdevelopment (host) is variable that captures the
nancial underdevelopment of the host country. To measure nancial underdevelopment, we have taken
a statistic for domestic credit to private sector relative to a countrys GDP, provided by the World
Banks World Development Indicators (WDI) database. We dene Financial underdevelopment (host)
as (
1orcsI.oCrcu.I
C1T
noa{
1orcsI.oCrcu.I
C1T
}) Weighted nancial underdevelopment (other locations)
is the lending-weighted nancial underdevelopment as dened in (24). Institutional weakness (host)
captures the institutional weakness of the host country. Also from the WDI database, to measure
institutional weakness, we use an index on the strength of investor protection. We dene the index
such that it ranges from 0 (strong investor protection) to 9 (weak investor protection). Accordingly,
weighted institutional weakness (other locations) is the lending-weighted institutional weakness as
dened in (24). Aliate-level productivity OP (host) captures the productivity of aliate i. We use
the method suggested by Olley and Pakes (1996) to estimate productivity. Weighted aliate-level
productivity OP (other locations) is lending-weighted productivity as dened in (24). Tangibility is
the aliate-specic xed-asset-to-total-asset ratio. Loss carryforward is a binary variable taking the
value one if an aliate reports a loss carryforward. Sales are the annual sales of an aliate in bill. e.
39
Table 2: Countries in Sample
Country Obs. Country Obs. Country Obs.
Albania 12 Haiti . Panama 99
Algeria 52 Honduras 37 Papua New Guinea .
Angola 9 Hong Kong 2,866 Paraguay 29
Argentina 1,303 Hungary 5,615 Peru 227
Armenia . Iceland 18 Philippines 497
Australia 3,577 India 2,082 Poland 8,754
Austria 12,580 Indonesia 755 Portugal 2,677
Azerbaijan 22 Iran 88 Qatar 6
Bahrain 18 Ireland 2,482 Romania 1,237
Bangladesh 53 Israel 300 Russian Federation 2,172
Belarus 50 Italy 11,909 Saudi Arabia 143
Belgium 6,301 Jamaica 8 Senegal 12
Bolivia 35 Japan 3,732 Sierra Leone .
Brazil 4,398 Jordan 11 Singapore 2,909
Bulgaria 513 Kazakhstan 87 Slovak Republic 1,707
Cameroon 33 Kenya 105 Slovenia 543
Canada 1,133 Korea, Rep. of 1,501 South Africa 2,470
Chile 791 Kuwait 12 Spain 11,206
China 6,437 Kyrgyzstan 8 Sri Lanka .
Columbia 459 Latvia 255 Swaziland .
Costa Rica 117 Lebanon 16 Sweden 3,821
Cte dIvoire 34 Lithuania 264 Switzerland 12,050
Croatia 592 Luxembourg 2,822 Syria 15
Czech Republic 7,592 Macedonia 51 Tanzania 24
Dem. Rep. Congo 7 Malawi 5 Thailand 1,002
Denmark 2,930 Malaysia 1,490 Trinidad & Tobago 31
Dominican Rep. . Mauritius 82 Tunisia 194
Ecuador 157 Mexico 2,566 Turkey 2,019
Egypt 393 Moldova 42 Uganda 22
El Salvador 59 Morocco 225 Ukraine 425
Estonia 235 Mozambique 6 Unit. A. Emirates 180
Ethiopia . Namibia . United Kingdom 17,968
Finland 1,336 Nepal . Uruguay 164
France 19,167 Netherlands 12,232 USA 28,756
Gabon 8 New Zealand 574 Venezuela 428
Gambia . Nicaragua 28 Vietnam 107
Georgia 10 Nigeria 150 Zambia 6
Ghana 25 Norway 949 Zimbabwe 20
Greece 1,392 Oman 12
Guatemala 105 Pakistan 184 All (118) Countries 227,558
Notes: Obs. refers to the total number of observations (aliates) from 1997 to 2007 in 118 host countries. .
denotes data, where reporting is not allowed due to condentiality reasons.
40
T
a
b
l
e
3
:
D
e
t
e
r
m
i
n
a
n
t
s
o
f
I
n
t
e
r
n
a
l
D
e
b
t

e
x
o
g
e
n
o
u
s
i
n
c
e
n
t
i
v
e
s
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
h
o
s
t
)
1
.
0
3
1
5
*
*
*
.
2
6
8
7
*
*
*
1
.
0
0
9
4
*
*
*
.
2
6
1
5
*
*
*
(
.
1
0
3
2
)
(
.
0
2
7
0
)
(
.
1
1
8
0
)
(
.
0
3
0
6
)
W
e
i
g
h
t
e
d
c
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
3
6
1
-
.
0
0
9
4
(
.
0
6
3
3
)
(
.
0
1
6
4
)
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
h
o
s
t
)
.
0
0
0
6
*
*
*
.
0
0
0
2
*
*
*
.
0
0
0
8
*
*
*
.
0
0
0
2
*
*
*
(
.
0
0
0
2
)
(
.
0
0
0
0
)
(
.
0
0
0
2
)
(
.
0
0
0
0
)
W
e
i
g
h
t
e
d

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
0
4
*
*
*
-
.
0
0
0
1
*
*
*
(
.
0
0
0
1
)
(
.
0
0
0
0
)
I
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
h
o
s
t
)
.
0
1
4
7
.
0
0
3
8
.
0
2
2
1
.
0
0
5
7
(
.
0
1
7
8
)
(
.
0
0
4
6
)
(
.
0
1
7
7
)
(
.
0
0
4
5
)
W
e
i
g
h
t
e
d
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
1
0
5
*
*
*
-
.
0
0
2
7
*
*
*
(
.
0
0
3
2
)
(
.
0
0
0
8
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
h
o
s
t
)
.
0
0
8
1
*
*
*
.
0
0
2
1
*
*
*
.
0
1
2
1
*
*
*
.
0
0
3
1
*
*
*
(
.
0
0
1
3
)
(
.
0
0
0
3
)
(
.
0
0
1
6
)
(
.
0
0
0
4
)
W
e
i
g
h
t
e
d
a

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
5
6
*
*
*
-
.
0
0
1
4
*
*
*
(
.
0
0
1
4
)
(
.
0
0
0
4
)
T
a
n
g
i
b
i
l
i
t
y
.
0
5
2
8
*
*
.
0
1
3
8
*
*
.
0
4
9
0
*
*
.
0
1
2
7
*
*
(
.
0
2
4
4
)
(
.
0
0
6
4
)
(
.
0
2
4
3
)
(
.
0
0
6
3
)
L
o
s
s
C
a
r
r
y
f
o
r
w
a
r
d
.
1
0
1
1
*
*
*
.
0
2
6
3
*
*
*
.
1
0
1
5
*
*
*
.
0
2
6
3
*
*
*
(
.
0
0
5
9
)
(
.
0
0
1
5
)
(
.
0
0
5
9
)
(
.
0
0
1
5
)
S
a
l
e
s
.
0
0
3
5
.
0
0
0
9
.
0
0
4
7
.
0
0
1
2
(
.
0
1
1
2
)
(
.
0
0
2
9
)
(
.
0
1
1
5
)
(
.
0
0
3
0
)
N
o
t
e
s
:
2
2
7
,
5
5
8
O
b
s
e
r
v
a
t
i
o
n
s
.
F
r
a
c
t
i
o
n
a
l
r
e
s
p
o
n
s
e
m
o
d
e
l
e
s
t
i
m
a
t
e
d
b
y
P
o
o
l
e
d
Q
M
L
E
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
r
e
f
e
r
s
t
o
i
n
t
e
r
n
a
l
c
r
o
s
s
-
b
o
r
d
e
r
d
e
b
t
.
A
l
l
r
e
g
r
e
s
s
i
o
n
s
i
n
c
l
u
d
e
t
i
m
e
d
u
m
m
i
e
s
a
n
d
a

l
i
a
t
e
-
s
p
e
c
i

x
e
d
e

e
c
t
s
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
.
T
h
e
s
t
a
n
d
a
r
d
e
r
r
o
r
s
f
o
r
t
h
e
A
P
E
s
w
e
r
e
o
b
t
a
i
n
e
d
u
s
i
n
g
b
o
o
t
s
t
r
a
p
p
i
n
g
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
41
Table 4: Endogenous incentives
Coe. APE
Corporate income tax (host) 3.5471*** .9176***
(.4650) (.1202)
Weighted corporate income tax (other locations) -2.9818*** -.7714***
(.5304) (.1370)
Financial underdevelopment (host) .0010** .0003**
(.0004) (.0001)
Weighted nancial underdevelopment (other locations) -.0007 -.0002
(.0005) (.0001)
Institutional weakness (host) .0625** .0162**
(.0275) (.0071)
Weighted institutional weakness (other locations) -.0653*** -.0169***
(.0233) (.0060)
Aliate-level productivity OP (host) .0088 .0023
(.0059) (.0015)
Weighted aliate-level productivity OP (other locations) -.0018 -.0005
(.0072) (.0019)
Tangibility .0531** .0137**
(.0242) (.0063)
Loss Carryforward .1019*** .0264***
(.0060) (.0015)
Sales .0107 .0028
(.0118) (.0030)
Notes: 227,558 Observations. Fractional response model estimated by Pooled QMLE. The dependent
variable refers to internal cross-border debt. All regressions include time dummies and aliate-specic
xed eects. Robust standard errors reported in parentheses and based on panel bootstrapping. *,
**, and *** indicate signicance at 10%, 5%, and 1%, respectively. Tests on the joint signicance of the
instruments yield the following robust F-statistics (in the order of the endogenous variables instrumented):
171.06, 219.17, 443.83, 618.25, respectively. Estimated coecients of the control function (powers of the
estimated rst-stage regression residuals): u
1
.I
: 3.0073*** (0.5404); u
2
.I
: .0004 (.0005); u
3
.I
: .0557**
(.0238); u
4
.I
: -.0030 (.0073).
42
Table 5: Allowing for wider (endogenous) incentives I
Coe. APE
Corporate income tax (host) 3.0949*** .8009***
(.4162) (.1077)
Weighted corporate income tax
1
(other locations) -2.4556*** -.6355***
(.4722) (.1221)
Financial underdevelopment (host) .0011*** .0003***
(.0004) (.0001)
Weighted nancial underdevelopment
1
(other locations) -.0008* -.0002*
(.0005) (.0001)
Institutional weakness (host) .0746*** .0193***
(.0259) (.0067)
Weighted institutional weakness
1
(other locations) -.0763*** -.0197***
(.0213) (.0055)
Aliate-level productivity OP (host) .0090 .0023
(.0058) (.0015)
Weighted aliate-level productivity
1
OP (other locations) -.0024 -.0006
(.0071) (.0018)
Tangibility .0533** .0138**
(.0242) (.0063)
Loss Carryforward .1015*** .0263***
(.0059) (.0015)
Sales .0099 .0026
(.0118) (.0031)
Notes: 227,558 Observations. Fractional response model estimated by Pooled QMLE. The dependent
variable refers to internal cross-border debt. All regressions include time dummies and aliate-specic
xed eects. Robust standard errors reported in parentheses and based on panel bootstrapping. *,
**, and *** indicate signicance at 10%, 5%, and 1%, respectively. Tests on the joint signicance
of the instruments yield the following robust F-statistics (in the order of the endogenous variables
instrumented): 276.89, 174.91, 612.16, 765.75, respectively. Estimated coecients of the control function
(powers of the estimated rst-stage regression residuals): u
1
.I
: 2.4570*** (0.4826); u
2
.I
: .0005 (.0005); u
3
.I
:
0.0666*** (0.0219); u
4
.I
: -.0023 (.0073). Note that the incentives arising from other locations are dened
such that dierentials between the host and other variables may become negative. All variables denoted
with 1 are dened according to Equation (25), where we apply the less strict condition o
,.I
= 1 if
i
.I
i
I
0.1o
,.
.
43
Table 6: Allowing for wider (endogenous) incentives II
Coe. APE
Corporate income tax (host) 2.1240*** .5501***
(.3718) (.0963)
Weighted corporate income tax
11
(other locations) -1.3198*** -.3418***
(.4178) (.1081)
Financial underdevelopment (host) .0010*** .0003***
(.0003) (.0001)
Weighted nancial underdevelopment
11
(other locations) -.0007* -.0002*
(.0004) (.0001)
Institutional weakness (host) .0824*** .0214***
(.0258) (.0067)
Weighted institutional weakness
11
(other locations) -.0854*** -.0221
(.0197) (.0051)
Aliate-level productivity OP (host) .0159*** .0041***
(.0057) (.0014)
Weighted aliate-level productivity
11
OP (other locations) -.0107 -.0028
(.0069) (.0017)
Tangibility .0518** .0134**
(.0243) (.0063)
Loss Carryforward .1010*** .0261***
(.0059) (.0015)
Sales .0061 .0016
(.0118) (.0031)
Notes: 227,558 Observations. Fractional response model estimated by Pooled QMLE. The dependent
variable refers to internal cross-border debt. All regressions include time dummies and aliate-specic
xed eects. Robust standard errors reported in parentheses and based on panel bootstrapping. *, **,
and *** indicate signicance at 10%, 5%, and 1%, respectively. Estimated coecients of the control func-
tion (powers of the estimated rst-stage regression residuals): u
1
.I
: 1.2584*** (.4291); u
2
.I
: .0005 (.0004);
u
3
.I
: .0778*** (.0201); u
4
.I
: .0068*** (.0073). All variables denoted with 11 are dened according to
Equation (25), where we apply the less strict condition o
,.I
= 1 if i
.I
i
I
0.5o
,.
.
44
Table 7: Allowing for wider (endogenous) incentives III
Coe. APE
Corporate income tax (host) 2.8036*** .7254***
(.4307) (.1113)
Weighted corporate income tax
111
(other locations) -2.0906*** -.5409***
(.4965) (.1284)
Financial underdevelopment (host) .0009** .0002**
(.0004) (.0001)
Weighted nancial underdevelopment
111
(other locations) -.0004 -.0001
(.0005) (.0001)
Institutional weakness (host) .0808*** .0209***
(.0152) (.0039)
Weighted institutional weakness
111
(other locations) -.1067*** -.0276***
(.0196) (.0050)
Aliate-level productivity OP (host) .0088 .0023
(.0056) (.0014)
Weighted aliate-level productivity
111
OP (other locations) -.0016 -.0004
(.0065) (.0017)
Tangibility .0561*** .0145***
(.0203) (.0053)
Loss Carryforward .1009*** .0261***
(.0064) (.0017)
Sales .0112 .0029
(.0098) (.0025)
Notes: 227,558 Observations. Fractional response model estimated by Pooled QMLE. The dependent
variable refers to internal cross-border debt. All regressions include time dummies and aliate-specic
xed eects. Robust standard errors reported in parentheses and based on panel bootstrapping. *,
**, and *** indicate signicance at 10%, 5%, and 1%, respectively. Tests on the joint signicance
of the instruments yield the following robust F-statistics (in the order of the endogenous variables
instrumented): 181.77, 187.38, 498.91, 735.22, respectively. Estimated coecients of the control function
(powers of the estimated rst-stage regression residuals): u
1
.I
: 2.1495*** (.5059); u
2
.I
: 0.0002 (0.0005);
u
3
.I
: 0.0958*** (0.0198); u
4
.I
: -0.0032 (0.0066). All variables denoted with 111 are dened according to
Equation (25), where we apply the less strict condition o
,.I
= 1 if i
.I
i
I
o
,.
.
45
T
a
b
l
e
8
:
A
l
t
e
r
n
a
t
i
v
e
p
r
o
d
u
c
t
i
v
i
t
y
m
e
a
s
u
r
e
s

e
n
d
o
g
e
n
o
u
s
i
n
c
e
n
t
i
v
e
s
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
h
o
s
t
)
3
.
3
9
6
5
*
*
*
0
.
8
8
1
1
*
*
*
3
.
3
4
2
6
*
*
*
.
8
6
3
8
*
*
*
(
.
4
3
2
9
)
(
.
1
1
2
2
)
(
.
4
4
3
5
)
(
.
1
1
4
4
)
W
e
i
g
h
t
e
d
c
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
2
.
7
9
8
0
*
*
*
-
.
7
2
5
8
*
*
*
-
2
.
7
0
3
9
*
*
*
-
.
6
9
8
8
*
*
*
(
.
4
9
3
0
)
(
.
1
2
7
7
)
(
.
5
2
5
3
)
(
.
1
3
5
6
)
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
h
o
s
t
)
.
0
0
0
9
*
*
.
0
0
0
2
*
*
.
0
0
0
8
*
.
0
0
0
2
*
(
.
0
0
0
4
)
(
.
0
0
0
1
)
(
.
0
0
0
4
)
(
.
0
0
0
1
)
W
e
i
g
h
t
e
d

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
0
5
-
.
0
0
0
1
-
.
0
0
0
4
-
.
0
0
0
1
(
.
0
0
0
5
)
(
.
0
0
0
1
)
(
.
0
0
0
5
)
(
.
0
0
0
1
)
I
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
h
o
s
t
)
.
0
6
5
8
*
*
.
0
1
7
1
*
*
.
0
6
9
1
*
*
.
0
1
7
8
7
*
*
(
.
0
2
7
5
)
(
.
0
0
7
1
)
(
.
0
2
8
6
)
(
.
0
0
7
4
)
W
e
i
g
h
t
e
d
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
7
1
1
*
*
*
-
.
0
1
8
4
*
*
*
-
.
0
7
4
9
*
*
*
-
.
0
1
9
3
*
*
*
(
.
0
2
3
1
)
(
.
0
0
6
0
)
(
.
0
2
3
6
)
(
.
0
0
6
1
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
L
e
v
P
e
t
(
h
o
s
t
)
-
.
0
0
7
4
-
.
0
0
1
9
(
.
0
0
8
4
)
(
.
0
0
2
2
)
W
e
i
g
h
t
e
d
a

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
L
e
v
P
e
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
.
0
2
3
9
.
0
0
6
2
(
.
0
1
8
0
)
(
.
0
0
4
7
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
A
l
t
(
h
o
s
t
)
.
0
0
5
2
.
0
0
1
4
(
.
0
0
7
1
)
(
.
0
0
1
8
)
W
e
i
g
h
t
e
d
a

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
A
l
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
6
0
-
.
0
0
1
6
(
.
0
0
7
4
)
(
.
0
0
1
9
)
T
a
n
g
i
b
i
l
i
t
y
.
0
5
5
6
*
*
.
0
1
4
4
*
*
.
0
4
9
1
*
*
.
0
1
2
7
*
*
(
.
0
2
4
9
)
(
.
0
0
6
5
)
(
.
0
2
5
0
)
(
.
0
0
6
5
)
L
o
s
s
C
a
r
r
y
f
o
r
w
a
r
d
.
1
0
2
4
.
0
2
6
5
.
1
0
2
9
*
*
*
.
0
2
6
6
*
*
*
(
.
0
0
6
0
)
(
.
0
0
1
6
)
(
.
0
0
5
8
)
(
.
0
0
1
5
)
S
a
l
e
s
.
0
1
7
2
.
0
0
4
4
.
0
1
6
7
2
.
0
0
4
3
(
.
0
1
1
4
)
(
.
0
0
2
9
)
(
.
0
1
0
7
)
(
.
0
0
2
8
)
N
o
t
e
s
:
R
e
s
u
l
t
s
r
e
p
o
r
t
e
d
i
n
t
h
e

r
s
t
a
n
d
s
e
c
o
n
d
(
t
h
i
r
d
a
n
d
f
o
u
r
t
h
)
c
o
l
u
m
n
s
b
a
s
e
d
o
n
2
2
7
,
5
5
8
(
2
1
7
,
1
0
3
)
o
b
s
e
r
v
a
t
i
o
n
s
.
F
r
a
c
t
i
o
n
a
l
r
e
s
p
o
n
s
e
m
o
d
e
l
e
s
t
i
m
a
t
e
d
b
y
P
o
o
l
e
d
Q
M
L
E
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
r
e
f
e
r
s
t
o
i
n
t
e
r
n
a
l
c
r
o
s
s
-
b
o
r
d
e
r
d
e
b
t
.
A
l
l
r
e
g
r
e
s
s
i
o
n
s
i
n
c
l
u
d
e
t
i
m
e
d
u
m
m
i
e
s
a
n
d
a

l
i
a
t
e
-
s
p
e
c
i

x
e
d
e

e
c
t
s
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
a
n
d
b
a
s
e
d
o
n
p
a
n
e
l
b
o
o
t
s
t
r
a
p
p
i
n
g
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
T
e
s
t
s
o
n
t
h
e
j
o
i
n
t
s
i
g
n
i

c
a
n
c
e
o
f
t
h
e
i
n
s
t
r
u
m
e
n
t
s
y
i
e
l
d
t
h
e
f
o
l
l
o
w
i
n
g
r
o
b
u
s
t
F
-
s
t
a
t
i
s
t
i
c
s
(
i
n
t
h
e
o
r
d
e
r
o
f
t
h
e
e
n
d
o
g
e
n
o
u
s
v
a
r
i
a
b
l
e
s
i
n
s
t
r
u
m
e
n
t
e
d
)
:
1
6
0
.
4
0
,
2
1
7
.
6
8
,
4
4
2
.
7
2
,
6
6
3
.
0
7
,
r
e
s
p
e
c
t
i
v
e
l
y
.
E
s
t
i
m
a
t
e
d
c
o
e

c
i
e
n
t
s
o
f
t
h
e
c
o
n
t
r
o
l
f
u
n
c
t
i
o
n
f
o
r
t
h
e
L
e
v
i
n
s
o
h
n
a
n
d
P
e
t
r
i
n
m
o
d
e
l
(
p
o
w
e
r
s
o
f
t
h
e
e
s
t
i
m
a
t
e
d

r
s
t
-
s
t
a
g
e
r
e
g
r
e
s
s
i
o
n
r
e
s
i
d
u
a
l
s
)
:
u
1 .
I
:
2
.
7
9
8
1
*
*
*
(
.
5
0
3
5
)
;
u
2 .
I
:
.
0
0
0
1
(
.
0
0
0
5
)
;
u
3 .
I
:
.
0
6
0
2
*
*
(
.
0
2
3
6
)
;
u
4 .
I
:
-
.
0
0
0
0
(
.
0
0
0
0
)
.
E
s
t
i
m
a
t
e
d
c
o
e

c
i
e
n
t
s
o
f
t
h
e
c
o
n
t
r
o
l
f
u
n
c
t
i
o
n
f
o
r
t
h
e
A
l
t
m
o
d
e
l
(
p
o
w
e
r
s
o
f
t
h
e
e
s
t
i
m
a
t
e
d

r
s
t
-
s
t
a
g
e
r
e
g
r
e
s
s
i
o
n
r
e
s
i
d
u
a
l
s
)
:
u
1 .
I
:
2
.
7
0
5
5
*
*
*
(
.
5
3
2
0
)
;
u
2 .
I
:
-
.
0
0
0
1
(
.
0
0
0
5
)
;
u
3 .
I
:
.
0
6
4
4
*
*
*
(
.
0
2
3
4
)
;
u
4 .
I
:
.
0
0
0
0
(
.
0
0
0
1
)
.
46
Table 9: Examples corresponding to Figure 4
Host Institutional Corporate Income Predicted Internal Marginal
Country Weakness (host) Tax Rate (host) Debt Ratio Eect of t
Bahamas 5.0000 .0000 .0228 .0172
China 5.2133 .3186 .1960 .0881
Greece 6.6600 .3221 .2258 .0957
Hong Kong .92000 .1661 .0479 .0318
Ireland 1.4000 .1125 .0340 .0240
Japan 2.7000 .4389 .2788 .1070
Mexico 5.3967 .3229 .2033 .0901
Singapore .39999 .2293 .0703 .0430
USA 1.4000 .4084 .2190 .0941
Notes: The predicted internal debt ratio (and the marginal tax eect in column 5) is evaluated at sample
means of the explanatory variables and the country-specic averages of the institutional weakness indicator
and the corporate income tax rate of the countries listed.
Table 10: Examples corresponding to Figure 5
Host Capital Market Corporate Income Predicted Internal Marginal
Country Underdevelopment Tax Rate Debt Ratio Eect of t
Bahamas 242.7126 .0000 .0206 .0158
China 208.8015 .3186 .1722 .0813
Greece 259.5229 .3221 .1889 .0861
Hong Kong 168.3227 .1661 .0634 .0396
Ireland 191.2044 .1125 .0451 .0303
Japan 124.8518 .4389 .2734 .1060
Mexico 300.3724 .3229 .2008 .0894
Singapore 209.5045 .2293 .1036 .0574
USA 140.3398 .4084 .2435 .0998
Notes: The predicted internal debt ratio (and the marginal tax eect in column 5) is evaluated at sample
means of the explanatory variables and the country-specic averages of the institutional weakness indicator
and the corporate income tax rate of the countries listed.
47
Online Appendix to:
Corporate Taxes and Internal Borrowing within
Multinational Firms
Peter Egger

, Christian Keuschnigg

, Valeria Merlo

, and Georg Wamser

September 18, 2012

Aliation: ETH Zrich, KOF, CEPR, CESifo, GEP Nottingham, and OUCBT. Address: ETH Zrich,
WEH E6, Weinbergstrasse 35, 8092 Zrich, Switzerland.

Aliation: University of St. Gallen, FGN-HSG, CEPR, and CESifo. Address: University of St. Gallen,
FGN-HSG, Varnbuelstrasse 19, 9000 St. Gallen, Switzerland.

Aliation: ETH Zrich. Address: ETH Zrich, WEH E7, Weinbergstrasse 35, 8092 Zrich, Switzerland.

Aliation: ETH Zrich. Address: ETH Zrich, WEH E8, Weinbergstrasse 35, 8092 Zrich, Switzerland.
A.1
Sensitivity analysis
The aim of this online appendix is to provide more detailed information on some of the sensi-
tivity checks mentioned but largely suppressed in the paper for reasons of space constraints.
In particular, those sensitivity checks relate to (i) the treatment of location choice of aliates
as to be endogenous, (ii) an alternative treatment of standard errors by using country-year
clustering instead of a panel bootstrap procedure, (iii) eects on small versus large rms,
(iv) the relative importance of conditioning on other fundamentals than corporate prot tax
rates, (v) the role of estimating nonlinear models (which respect to the boundedness of the
dependent variable) relative to linear (OLS) models, (vi) the consideration of interaction
terms of the fundamental variables so that their nonlinear eects on internal borrowing do
not only accrue to the nonlinear functional form of the estimator for fractional dependent
variables, and (vii) the consistency of internal borrowing within a country with rm-level
productivity dierences across all units of a rm in that country and year. We discuss any
one of those considerations in more detail below.
(i) A control function approach to endogenous aliate location choice
We address the potential endogeneity of location choice by adopting a semi-parametric con-
trol function approach, akin to the one utilized with binary choice problems. For this, we
estimate annual conditional logit location choice models based on the following determinants
of (potential) aliate location (with data sources added in parentheses): the statutory cor-
porate tax rate in each potential host country (as in the paper); nancial underdevelopment
in each potential host country (as in the paper); institutional weakness in each potential
host country (as in the paper); log real GDP at constant U.S. dollars of the year 2000 in
each potential host country (World Banks World Development Indicators 2010); the cor-
ruption perception index for each potential host country (Transparency International); the
investment freedom index for each potential host country (Heritage Foundations Heritage In-
dicators); the costs of starting a business in each potential host country (World Banks World
Development Indicators 2010); log bilateral distance between Germany and each (potential)
host country (Centre dtudes Prospectives et dInformations Internationales Geographical
Database).
Tables A.1 and A.2
We report the conditional logit parameter estimates for each covered year between 1996
and 2007 in Table A.1, below. The table indicates that the parameters are not stable over
time. However, we should be careful with over-emphasizing this result since the conditional
logit model is nonlinear and the parameters should not be directly compared with each
other. We utilize the linear prediction from the location choice models across all the years
in original form and, at the same time, in squared form. These two variables (together
with the parameters that have to be estimated on them) form a quadratic control function.
This function is added to the specications in the last two columns of Table 4 of the paper
(assuming exogeneity of the potential lending weighting of other aliates) and the one in
A.2
Table 5 of the paper (assuming endogeneity of the weighting). This control function approach
is, e.g., similar in spirit to the semiparametric control function approach adopted to selection
into export markets in Helpman, Melitz, and Rubinstein (2008). The corresponding results
are summarized in Table A.2, and they suggest that controlling for endogenous location
choice in the proposed way does not lead to any big qualitative change relative to the
benchmark results in Tables 4 and 5 of the paper. In fact, the (robust) standard errors on
the parameters of the control function suggest that there is no bias in the parameters in
Tables 4 and 5 from endogenous selection into locations by the multinational rm.
(ii) Country-year clustering of standard errors instead of panel-boot-
strapped standard errors
We report counterparts to the last two columns of Table 4 and to Table 5 using country-year
clustering instead of panel bootsrapping in Table A.3, below.
Table A.3
It turns out that controlling for country-year clusters in the variance-covariance matrix leads
to slightly bigger standard errors than panel bootstrapping. However, there is no qualitative
change in the conclusions.
(iii) Regression results for small versus large rms
To assess the potential relevance of rm size (and selectivity bias in previous research) for
the results, we run the regressions in the last pair of columns of Table 4 and in Table 5 for
two alternative subsamples: one that contains rms that are smaller than the rm at the
25th centile and one that contains rms that are larger than the rm at the 75th centile. The
corresponding results are summarized in Table A.4, below, and they should be compared
with the benchmark results in Tables 4 and 5.
Table A.4
If anything, the Table A.4 suggests that utilizing a subsample of large rms leads to somewhat
smaller coecients and average partial eects than utilizing a subsample of small rms.
Hence, we conclude that to some (though not an overwhelming) extent the dierence
in magnitude of tax eects for internal borrowing between this study and previous research
may accrue to the selective reliance on larger rms in several of the studies.
(iv) Regression results without the three non-tax fundamentals
We report counterparts to the last two columns of Table 4 and to Table 5 disregarding the
three non-tax fundamentals in Table A.5, below.
Table A.5
A.3
It turns out that ignoring the three non-tax fundamentals from the specication leads to
somewhat bigger parameters for the corporate tax variables than in the benchmark spec-
ications in Tables 4 and 5. The latter holds true no matter of whether one controls for
weighted corporate tax rates elsewhere in the rm or not. This suggests that the smaller ef-
fects found in previous research are unlikely due to the omission of those other fundamentals
from the corresponding model specications.
(v) Regression results for linear (OLS) models
We report counterparts to the last two columns of Table 4 and to Table 5 based on (linear)
xed eects and two-stage least squares xed eects in Table A.6, below.
Table A.6
Indeed, it turns out that assuming a log-linear regression model (which is inadequate since it
does not pay attention to the boundedness of the dependent variable) leads to much smaller
coecients and average partial eects of the statutory income tax variable. To see this,
compare the rst column of Table A.6 with its nonlinear counterpart in the last two columns
of Table 4. It turns out that the corporate tax parameter (and the average partial eect)
is much smaller in the linear model with exogenous weighting than in the nonlinear model
as reported in the last two columns of Table 4. The parameters of the tax variables even
turn (and take on implausible signs) in case of two-stage least-squares estimation relative to
the nonlinear two-stage least-squares model in Table 5. This suggests that the inadequate
linear modeling of internal lending (in levels or as a ratio) is identied as one potentially
fundamental problem with the data at hand.
(vi) Estimating nonlinear models with interaction terms of the fun-
damentals
We report counterparts to the last two columns of Table 4 and to Table 5 based on speci-
cations which include interactive terms between all four fundamental variables employed in
Tables 4 and 5 in Table A.7, below. This is meant to account for a nonlinear impact of the
fundamentals which goes beyond the nonlinearity as imposed by the functional form of the
nonlinear models estimated. Hence, in the corresponding models there are now two sources
of nonlinearity: one that relates to the specication of the latent process and the other one
relates to the nonlinear mapping of the latent process onto the unit space.
Table A.7
The corresponding parameter estimates are impossible to interpret since the fractional re-
sponse model is nonlinear per se. However, the results can be described as follows. First,
the predictions of the surfaces contain somewhat more extreme values (zero debt ratios at
the lower bounds and unitary debt ratios at the upper bounds) than Figures 4-6. However,
the qualitative shapes of the - relationship counterpart to Figure 4, the - relationship
A.4
counterpart to Figure 5, and the -
w
relationship counterpart to Figure 6 is the same. While
the extreme values (zeros and ones of outcome) are mostly outside of the region supported
by the data anyway, the slope of the estimated surfaces are steeper in the support region.
Hence, if anything, including interaction terms among the fundamentals renders the debt
ratio even more responsive to the fundamentals (including tax rates) than in Figures 4-6.
(vii) Consistency of internal borrowing within a country with rm-
level productivity dierences
We report counterparts to the last two columns of Table 4 and to Table 5 based on speci-
cations which discern between the average productivity within a country (and year) and the
deviation from the latter per aliate in Tables A.8 and A.9. For this, we split the (measured
or estimated) rm-level productivity variables into two orthogonal components. Let us de-
note a generic productivity variable for aliate i in year t by
it
as in the paper, and the
number of all units of the rm aliate i belongs to in the same country as i and in year t by
B
it
. Then, the average productivity of that rm in the country where i is located and year
t may be denoted as

it
=
1
B
it

B
it
i=1

it
, and the deviation of i from this average in year t is

it
=
it

it
. While the regressions in Tables 4-8 included
it
as a regressor, the alternative
ones as counterparts to the benchmark regressions on the outer right of Table 4 and in Table
5 employ

it
and

it
as two separate regressors. Obviously,
it
=

it
for all rms which hold
only a single aliate in the country where i is hosted in year t. Hence, identication of the
separate eects of

it
and

it
comes entirely from multi-unit rms in a single country and
year.
Tables A.8 and A.9
Focusing on those results where the productivity measures carry statistically signicant
coecients, the ndings suggest that

it
and

it
enter with the same sign and have coecients
of quite similar magnitude. Hence, variability of productivity across units in the same rm,
host country, and year is as important for triggering internal lending as the variability of
average productivity across host countries is in the same rm and year.
Tables
A.5
T
a
b
l
e
A
.
1
:
L
o
c
a
t
i
o
n
D
e
c
i
s
i
o
n
1
9
9
6
1
9
9
7
1
9
9
8
1
9
9
9
2
0
0
0
2
0
0
1
2
0
0
2
2
0
0
3
2
0
0
4
2
0
0
5
2
0
0
6
2
0
0
7
C
o
r
p
o
r
a
t
e
-
2
.
0
0
1
1
*
*
*
-
1
.
5
8
9
5
*
*
*
-
2
.
4
5
8
9
*
*
*
-
.
2
9
3
2
.
1
2
5
0
-
1
.
0
7
9
2
*
*
*
-
.
1
0
9
4
.
1
1
4
5
-
1
.
5
1
5
1
*
*
*
-
2
.
3
2
4
6
*
*
*
-
1
.
7
7
6
6
*
*
*
-
1
.
9
5
5
0
*
*
*
i
n
c
o
m
e
t
a
x
(
.
1
7
2
6
)
(
.
1
5
5
2
)
(
.
1
9
0
4
)
(
.
1
8
7
9
)
(
.
1
9
5
7
)
(
.
1
7
3
5
)
(
.
1
5
2
7
)
(
.
1
6
3
7
)
(
.
1
5
4
3
)
(
.
1
5
3
0
)
(
.
1
4
4
1
)
(
.
1
5
1
8
)
F
i
n
a
n
c
i
a
l
-
.
0
0
5
8
*
*
*
-
.
0
0
5
5
*
*
*
-
.
0
0
2
7
*
*
*
-
.
0
0
3
4
*
*
*
-
.
0
0
1
9
*
*
*
-
.
0
0
0
3
-
.
0
0
0
3
-
.
0
0
0
4
-
.
0
0
0
7
*
*
-
.
0
0
0
0
3
.
0
0
1
0
*
*
*
.
0
0
2
9
*
*
*
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
.
0
0
0
4
)
(
.
0
0
0
3
)
(
.
0
0
0
4
)
(
.
0
0
0
3
)
(
.
0
0
0
3
)
(
.
0
0
0
3
)
(
.
0
0
0
3
)
(
.
0
0
0
3
)
(
.
0
0
0
3
)
(
.
0
0
0
3
)
(
.
0
0
0
2
)
(
.
0
0
0
2
)
I
n
s
t
i
t
u
t
i
o
n
a
l
.
1
0
8
1
*
*
*
.
1
2
7
9
*
*
*
.
0
4
6
0
*
*
*
.
1
0
3
4
*
*
*
.
0
8
6
5
*
*
*
.
0
7
9
0
*
*
*
.
0
6
1
5
*
*
*
.
0
6
0
5
*
*
*
.
0
6
4
3
*
*
*
.
0
7
5
2
*
*
*
.
0
7
6
4
*
*
*
.
1
0
0
0
*
*
*
w
e
a
k
n
e
s
s
(
.
0
0
9
2
)
(
.
0
0
8
0
)
(
.
0
1
0
0
)
(
.
0
0
7
0
)
(
.
0
0
7
2
)
(
.
0
0
6
5
)
(
.
0
0
6
2
)
(
.
0
0
6
3
)
(
.
0
0
6
4
)
(
.
0
0
6
2
)
(
.
0
0
6
4
)
(
.
0
0
7
0
)
l
o
g
.
7
9
8
6
*
*
*
.
7
8
0
6
*
*
*
.
8
5
3
1
*
*
*
.
7
9
9
6
*
*
*
.
7
9
3
1
*
*
*
.
8
4
4
9
*
*
*
.
8
3
1
6
*
*
*
.
8
3
4
0
*
*
*
.
8
4
4
3
*
*
*
.
8
6
7
2
*
*
*
.
8
6
8
7
*
*
*
.
8
8
2
9
*
*
*
G
D
P
(
.
0
1
2
4
)
(
.
0
1
1
3
)
(
.
0
1
1
1
)
(
.
0
0
8
9
)
(
.
0
0
8
5
)
(
.
0
0
9
2
)
(
.
0
0
8
3
)
(
.
0
0
8
3
)
(
.
0
0
8
3
)
(
.
0
0
8
2
)
(
.
0
0
7
7
)
(
.
0
0
8
7
)
C
o
r
r
u
p
t
i
o
n
.
0
4
2
2
*
*
*
.
0
2
7
9
*
*
*
.
0
1
9
9
*
*
.
0
7
1
4
*
*
*
.
0
7
7
1
*
*
*
.
0
8
1
3
*
*
*
.
0
5
5
6
*
*
*
.
0
5
6
0
*
*
*
.
0
6
4
8
*
*
*
.
0
7
2
4
*
*
*
.
1
1
5
2
*
*
*
.
1
4
9
3
*
*
*
P
e
r
c
e
p
t
i
o
n
(
.
0
0
8
0
)
(
.
0
0
7
0
)
(
.
0
0
7
8
)
(
.
0
0
6
4
)
(
.
0
0
5
6
)
(
.
0
0
6
8
)
(
.
0
0
7
5
)
(
.
0
0
7
6
)
(
.
0
0
7
9
)
(
.
0
0
7
9
)
(
.
0
0
7
0
)
(
.
0
0
6
4
)
I
n
v
e
s
t
m
e
n
t
.
0
1
7
7
*
*
*
.
0
2
5
4
*
*
*
.
0
3
4
3
*
*
*
.
0
2
2
7
*
*
*
.
0
2
1
8
*
*
*
.
0
1
8
3
*
*
*
.
0
1
8
0
*
*
*
.
0
1
8
4
*
*
*
.
0
1
9
3
*
*
*
.
0
1
6
1
*
*
*
.
0
1
1
1
*
*
*
.
0
1
4
1
*
*
*
F
r
e
e
d
o
m
(
.
0
0
1
4
)
(
.
0
0
1
3
)
(
.
0
0
1
5
)
(
.
0
0
1
1
)
(
.
0
0
1
0
)
(
.
0
0
0
7
)
(
.
0
0
0
6
)
(
.
0
0
0
6
)
(
.
0
0
0
7
)
(
.
0
0
0
7
)
(
.
0
0
0
6
)
(
.
0
0
0
7
)
C
o
s
t
o
f
S
t
a
r
t
i
n
g
-
.
0
0
9
7
*
*
*
-
.
0
1
0
3
*
*
*
-
.
0
0
5
1
*
*
*
-
.
0
0
6
0
*
*
*
-
.
0
0
7
4
*
*
*
-
.
0
0
7
9
*
*
*
-
.
0
0
9
5
*
*
*
-
.
0
0
9
1
*
*
*
-
.
0
0
5
4
*
*
*
-
.
0
0
5
3
*
*
*
-
.
0
0
6
9
*
*
*
-
.
0
0
7
1
*
*
*
a
B
u
s
i
n
e
s
s
(
.
0
0
1
1
)
(
.
0
0
1
0
)
(
.
0
0
0
7
)
(
.
0
0
0
8
)
(
.
0
0
0
8
)
(
.
0
0
0
7
)
(
.
0
0
0
7
)
(
.
0
0
0
7
)
(
.
0
0
0
6
)
(
.
0
0
0
6
)
(
.
0
0
0
7
)
(
.
0
0
0
8
)
l
o
g
-
.
6
2
8
2
*
*
*
-
.
6
0
7
2
*
*
*
-
.
5
2
2
6
*
*
*
-
.
5
9
4
1
*
*
*
-
.
5
6
1
5
*
*
*
-
.
5
2
5
0
*
*
*
-
.
5
6
9
2
*
*
*
-
.
5
9
1
3
*
*
*
-
.
5
4
6
4
*
*
*
-
.
4
8
7
3
*
*
*
-
.
5
2
4
9
*
*
*
-
.
4
8
6
7
*
*
*
D
i
s
t
a
n
c
e
(
.
0
1
4
7
)
(
.
0
1
3
0
)
(
.
0
1
4
7
)
(
.
0
1
1
8
)
(
.
0
0
9
7
)
(
.
0
0
9
5
)
(
.
0
0
9
8
)
(
.
0
1
0
6
)
(
.
0
1
0
0
)
(
.
0
0
9
3
)
(
.
0
0
9
1
)
(
.
0
0
8
9
)
O
b
s
e
r
v
a
t
i
o
n
s
5
7
6
,
9
6
0
6
5
3
,
5
2
0
7
3
5
,
1
9
5
1
,
3
8
8
,
8
3
9
1
,
4
0
8
,
3
5
0
1
,
6
6
4
,
7
9
6
1
,
7
4
7
,
1
7
0
1
,
9
6
8
,
7
0
2
1
,
9
8
4
,
1
9
2
2
,
2
3
5
,
3
3
7
2
,
3
1
8
,
1
9
0
2
,
2
8
7
,
5
0
0
N
o
t
e
s
:
C
o
n
d
i
t
i
o
n
a
l
l
o
g
i
t
e
s
t
i
m
a
t
e
s
o
f
t
h
e
l
o
c
a
t
i
o
n
d
e
c
i
s
i
o
n
o
f
G
e
r
m
a
n
m
u
l
t
i
n
a
t
i
o
n
a
l
s
f
r
o
m
1
9
9
6
t
o
2
0
0
7
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
C
o
n
t
r
o
l
v
a
r
i
a
b
l
e
s
a
r
e
t
a
k
e
n
f
r
o
m
d
i

e
r
e
n
t
s
o
u
r
c
e
s
.
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
i
s
t
h
e
s
t
a
t
u
t
o
r
y
t
a
x
r
a
t
e
o
f
a
h
o
s
t
c
o
u
n
t
r
y
.
T
h
e
t
a
x
d
a
t
a
i
s
c
o
l
l
e
c
t
e
d
f
r
o
m
d
a
t
a
b
a
s
e
s
p
r
o
v
i
d
e
d
b
y
t
h
e
I
n
t
e
r
n
a
t
i
o
n
a
l
B
u
r
e
a
u
o
f
F
i
s
c
a
l
D
o
c
u
m
e
n
t
a
t
i
o
n
(
I
B
F
D
)
a
n
d
t
a
x
s
u
r
v
e
y
s
p
r
o
v
i
d
e
d
b
y
E
r
n
s
t
&
Y
o
u
n
g
,
P
w
C
,
a
n
d
K
P
M
G
.
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
i
s
a
v
a
r
i
a
b
l
e
t
h
a
t
c
a
p
t
u
r
e
s
t
h
e

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
o
f
t
h
e
h
o
s
t
c
o
u
n
t
r
y
.
T
o
m
e
a
s
u
r
e

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
,
w
e
h
a
v
e
t
a
k
e
n
a
s
t
a
t
i
s
t
i
c
f
o
r
d
o
m
e
s
t
i
c
c
r
e
d
i
t
t
o
p
r
i
v
a
t
e
s
e
c
t
o
r
r
e
l
a
t
i
v
e
t
o
a
c
o
u
n
t
r
y

s
G
D
P
,
p
r
o
v
i
d
e
d
b
y
t
h
e
W
o
r
l
d
B
a
n
k

s
W
o
r
l
d
D
e
v
e
l
o
p
m
e
n
t
I
n
d
i
c
a
t
o
r
s
(
W
D
I
)
d
a
t
a
b
a
s
e
.
W
e
t
h
e
n
d
e

n
e
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
a
s
(
|
D
o
m
e
s
t
i
c
C
r
e
d
i
t
G
D
P

m
a
x
{
D
o
m
e
s
t
i
c
C
r
e
d
i
t
G
D
P
}
|
)
.
I
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
h
o
s
t
)
c
a
p
t
u
r
e
s
t
h
e
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
o
f
t
h
e
h
o
s
t
c
o
u
n
t
r
y
.
A
l
s
o
f
r
o
m
t
h
e
W
D
I
d
a
t
a
b
a
s
e
,
t
o
m
e
a
s
u
r
e
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
,
w
e
u
s
e
a
n
i
n
d
e
x
o
n
t
h
e
s
t
r
e
n
g
t
h
o
f
i
n
v
e
s
t
o
r
p
r
o
t
e
c
t
i
o
n
.
W
e
d
e

n
e
t
h
e
i
n
d
e
x
s
u
c
h
t
h
a
t
i
t
r
a
n
g
e
s
f
r
o
m
0
(
s
t
r
o
n
g
i
n
v
e
s
t
o
r
p
r
o
t
e
c
t
i
o
n
)
t
o
9
(
w
e
a
k
i
n
v
e
s
t
o
r
p
r
o
t
e
c
t
i
o
n
)
.
l
o
g
G
D
P
m
e
a
s
u
r
e
s
t
h
e
r
e
a
l
G
D
P
a
t
c
o
n
s
t
a
n
t
U
.
S
.
d
o
l
l
a
r
s
o
f
t
h
e
y
e
a
r
2
0
0
0
a
n
d
i
s
t
a
k
e
n
f
r
o
m
t
h
e
W
o
r
l
d
B
a
n
k

s
W
o
r
l
d
D
e
v
e
l
o
p
m
e
n
t
I
n
d
i
c
a
t
o
r
s
2
0
0
9
.
C
P
I
(
C
o
r
r
u
p
t
i
o
n
P
e
r
c
e
p
t
i
o
n
I
n
d
e
x
)
i
s
p
u
b
l
i
s
h
e
d
a
n
n
u
a
l
l
y
b
y
T
r
a
n
s
p
a
r
e
n
c
y
I
n
t
e
r
n
a
t
i
o
n
a
l
.
I
t
r
a
n
k
s
c
o
u
n
t
r
i
e
s
i
n
t
e
r
m
s
o
f
p
e
r
c
e
i
v
e
d
l
e
v
e
l
s
o
f
c
o
r
r
u
p
t
i
o
n
,
a
s
d
e
t
e
r
m
i
n
e
d
b
y
e
x
p
e
r
t
a
s
s
e
s
s
m
e
n
t
s
a
n
d
o
p
i
n
i
o
n
s
u
r
v
e
y
s
.
T
h
e
s
c
o
r
e
s
r
a
n
g
e
f
r
o
m
1
0
(
c
o
u
n
t
r
y
p
e
r
c
e
i
v
e
d
a
s
v
i
r
t
u
a
l
l
y
c
o
r
r
u
p
t
i
o
n
f
r
e
e
)
t
o
0
(
c
o
u
n
t
r
y
p
e
r
c
e
i
v
e
d
a
s
a
l
m
o
s
t
t
o
t
a
l
l
y
c
o
r
r
u
p
t
)
.
T
h
e
i
n
v
e
s
t
m
e
n
t
f
r
e
e
d
o
m
i
n
d
e
x
I
n
v
e
s
t
m
e
n
t
F
r
e
e
d
o
m
i
s
t
a
k
e
n
f
r
o
m
t
h
e
H
e
r
i
t
a
g
e
I
n
d
i
c
a
t
o
r
s
d
a
t
a
b
a
s
e
.
T
h
e
i
n
d
e
x
c
a
n
t
a
k
e
o
n
v
a
l
u
e
s
b
e
t
w
e
e
n
0
a
n
d
1
0
0
;
h
i
g
h
e
r
v
a
l
u
e
s
a
r
e
a
s
s
o
c
i
a
t
e
d
w
i
t
h
m
o
r
e
i
n
v
e
s
t
m
e
n
t
f
r
e
e
d
o
m
.
C
o
s
t
o
f
S
t
a
r
t
i
n
g
a
B
u
s
i
n
e
s
s
i
s
f
r
o
m
W
o
r
l
d
B
a
n
k

s
D
o
i
n
g
B
u
s
i
n
e
s
s
D
a
t
a
b
a
s
e
a
n
d
m
e
a
s
u
r
e
s
t
h
e
c
o
s
t
o
f
s
t
a
r
t
i
n
g
a
b
u
s
i
n
e
s
s
r
e
l
a
t
i
v
e
t
o
i
n
c
o
m
e
p
e
r
c
a
p
i
t
a
.
l
o
g
D
i
s
t
a
n
c
e
i
s
t
h
e
l
o
g
o
f
t
h
e
d
i
s
t
a
n
c
e
(
i
n
k
i
l
o
m
e
t
e
r
)
b
e
t
w
e
e
n
t
h
e
m
o
s
t
p
o
p
u
l
a
t
e
d
c
i
t
i
e
s
i
n
t
h
e
h
o
s
t
c
o
u
n
t
r
y
a
n
d
t
h
e
p
a
r
e
n
t
c
o
u
n
t
r
y
(
G
e
r
m
a
n
y
)
.
T
h
e
d
i
s
t
a
n
c
e
v
a
r
i
a
b
l
e
i
s
t
a
k
e
n
f
r
o
m
t
h
e
C
e
n
t
r
e
d

t
u
d
e
s
P
r
o
s
p
e
c
t
i
v
e
s
e
t
d

I
n
f
o
r
m
a
t
i
o
n
s
I
n
t
e
r
n
a
t
i
o
n
a
l
e
s
.
A.6
T
a
b
l
e
A
.
2
:
D
e
t
e
r
m
i
n
a
n
t
s
o
f
I
n
t
e
r
n
a
l
D
e
b
t

E
x
o
g
e
n
o
u
s
a
n
d
E
n
d
o
g
e
n
o
u
s
I
n
c
e
n
t
i
v
e
s
U
s
i
n
g
a
Q
u
a
d
r
a
t
i
c
C
o
n
t
r
o
l
F
u
n
c
t
i
o
n
f
o
r
E
n
d
o
g
e
n
o
u
s
L
o
c
a
t
i
o
n
C
h
o
i
c
e
E
x
o
g
.
i
n
c
e
n
t
i
v
e
s
E
n
d
o
g
.
i
n
c
e
n
t
i
v
e
s
(
c
o
m
p
.
t
o
T
a
b
l
e
4
)
(
c
o
m
p
.
t
o
T
a
b
l
e
5
)
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
h
o
s
t
)
.
9
5
3
0
*
*
*
.
2
4
9
4
*
*
*
3
.
3
5
8
8
*
*
*
.
8
7
7
9
*
*
*
(
.
1
1
3
5
)
(
.
0
2
9
7
)
(
.
3
8
0
3
)
(
.
0
9
9
1
)
W
e
i
g
h
t
e
d
c
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
3
1
2
-
.
0
0
8
2
-
2
.
8
1
6
1
*
*
*
-
.
7
3
6
1
*
*
*
(
.
0
6
5
5
)
(
.
0
1
7
1
)
(
.
4
4
2
7
)
(
.
1
1
5
4
)
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
h
o
s
t
)
.
0
0
0
7
*
*
*
.
0
0
0
2
*
*
*
.
0
0
1
2
*
*
.
0
0
0
3
*
*
(
.
0
0
0
2
)
(
.
0
0
0
1
)
(
.
0
0
0
5
)
(
.
0
0
0
1
)
W
e
i
g
h
t
e
d

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
0
4
*
*
*
-
.
0
0
0
1
*
*
*
-
.
0
0
1
1
*
-
.
0
0
0
3
*
(
.
0
0
0
1
)
(
.
0
0
0
0
)
(
.
0
0
0
6
)
(
.
0
0
0
2
)
I
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
h
o
s
t
)
.
0
1
8
6
.
0
0
4
9
.
0
5
1
2
*
.
0
1
3
4
*
(
.
0
1
8
6
)
(
.
0
0
4
9
)
(
.
0
2
6
4
)
(
.
0
0
6
9
)
W
e
i
g
h
t
e
d
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
1
1
2
*
*
*
-
.
0
0
2
9
*
*
*
-
.
0
5
4
2
*
*
-
.
0
1
4
2
*
*
(
.
0
0
3
2
)
(
.
0
0
0
8
)
(
.
0
2
4
3
)
(
.
0
0
6
4
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
h
o
s
t
)
.
0
1
2
0
*
*
*
.
0
0
3
1
*
*
*
.
0
0
6
7
.
0
0
1
7
(
.
0
0
1
9
)
(
.
0
0
0
5
)
(
.
0
0
5
6
)
(
.
0
0
1
5
)
W
e
i
g
h
t
e
d
a

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
5
6
*
*
*
-
.
0
0
1
4
*
*
*
.
0
0
0
8
.
0
0
0
2
(
.
0
0
1
4
)
(
.
0
0
0
4
)
(
.
0
0
6
6
)
(
.
0
0
1
7
)
T
a
n
g
i
b
i
l
i
t
y
.
0
4
6
5
*
.
0
1
2
1
*
.
0
5
0
6
*
*
.
0
1
3
7
*
*
(
.
0
2
6
0
)
(
.
0
0
6
8
)
(
.
0
2
6
1
)
(
.
0
0
6
3
)
L
o
s
s
c
a
r
r
y
f
o
r
w
a
r
d
.
0
9
8
4
*
*
*
.
0
2
5
8
*
*
*
.
0
9
8
7
*
*
*
.
0
2
6
4
*
*
*
(
.
0
0
5
7
)
(
.
0
0
1
5
)
(
.
0
0
5
7
)
(
.
0
0
1
5
)
S
a
l
e
s
.
0
0
4
6
.
0
0
0
9
.
0
0
8
7
.
0
0
2
8
(
.
0
1
2
4
)
(
.
0
0
3
3
)
(
.
0
1
2
6
)
(
.
0
0
3
0
)
C
o
n
t
r
o
l
f
u
n
c
t
i
o
n
:
L
i
n
e
a
r
i
n
d
e
x
f
r
o
m
c
o
n
d
i
t
i
o
n
a
l
l
o
g
i
t
i
n
T
a
b
l
e
A
.
1
-
.
0
2
2
0
-
.
0
0
5
7
-
.
0
1
4
3
-
.
0
0
3
7
(
.
0
3
1
7
)
(
.
0
0
8
3
)
(
.
0
3
2
1
)
(
.
0
0
8
4
)
C
o
n
t
r
o
l
f
u
n
c
t
i
o
n
:
L
i
n
e
a
r
i
n
d
e
x
s
q
u
a
r
e
d
f
r
o
m
c
o
n
d
i
t
i
o
n
a
l
l
o
g
i
t
i
n
T
a
b
l
e
A
.
1
.
0
0
0
5
.
0
0
0
1
.
0
0
0
5
.
0
0
0
1
(
.
0
0
0
8
)
(
.
0
0
0
2
)
(
.
0
0
0
8
)
(
.
0
0
0
2
)
N
o
t
e
s
:
2
2
7
,
5
5
8
o
b
s
e
r
v
a
t
i
o
n
s
.
F
r
a
c
t
i
o
n
a
l
r
e
s
p
o
n
s
e
m
o
d
e
l
e
s
t
i
m
a
t
e
d
b
y
P
o
o
l
e
d
Q
M
L
E
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
r
e
f
e
r
s
t
o
i
n
t
e
r
n
a
l
c
r
o
s
s
-
b
o
r
d
e
r
d
e
b
t
.
A
l
l
r
e
g
r
e
s
s
i
o
n
s
i
n
c
l
u
d
e
t
i
m
e
d
u
m
m
i
e
s
a
n
d
a

l
i
a
t
e
-
s
p
e
c
i

x
e
d
e

e
c
t
s
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
a
n
d
b
a
s
e
d
o
n
p
a
n
e
l
b
o
o
t
s
t
r
a
p
p
i
n
g
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
A.7
T
a
b
l
e
A
.
3
:
D
e
t
e
r
m
i
n
a
n
t
s
o
f
I
n
t
e
r
n
a
l
D
e
b
t

E
x
o
g
e
n
o
u
s
a
n
d
E
n
d
o
g
e
n
o
u
s
I
n
c
e
n
t
i
v
e
s
w
i
t
h
C
o
u
n
t
r
y
-
Y
e
a
r
C
l
u
s
t
e
r
e
d
R
a
t
h
e
r
T
h
a
n
P
a
n
e
l
-
b
o
o
t
s
t
r
a
p
p
e
d
S
t
a
n
d
a
r
d
E
r
r
o
r
s
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
h
o
s
t
)
1
.
0
0
9
4
*
*
*
.
2
6
1
5
*
*
*
3
.
5
4
7
1
*
*
*
.
9
1
7
6
*
*
*
(
.
1
9
9
3
)
(
.
0
5
1
6
)
(
.
3
8
1
8
)
(
.
0
9
8
8
)
W
e
i
g
h
t
e
d
c
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
3
6
1
-
.
0
0
9
4
-
2
.
9
8
1
8
*
*
*
-
.
7
7
1
4
*
*
*
(
.
0
9
9
9
)
(
.
0
2
5
9
)
(
.
4
1
3
8
)
(
.
1
0
7
0
)
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
h
o
s
t
)
.
0
0
0
8
*
*
.
0
0
0
2
*
*
.
0
0
1
0
*
*
.
0
0
0
3
*
*
(
.
0
0
0
3
)
(
.
0
0
0
1
)
(
.
0
0
0
4
)
(
.
0
0
0
1
)
W
e
i
g
h
t
e
d

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
0
4
*
*
-
.
0
0
0
1
*
*
-
.
0
0
0
7
*
-
.
0
0
0
2
*
(
.
0
0
0
1
)
(
.
0
0
0
0
4
)
(
.
0
0
0
4
)
(
.
0
0
0
1
)
I
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
h
o
s
t
)
.
0
2
2
1
.
0
0
5
7
.
0
6
2
5
.
0
1
6
2
(
.
0
5
8
8
)
(
.
0
1
5
2
)
(
.
0
5
5
0
)
(
.
0
1
4
2
)
W
e
i
g
h
t
e
d
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
1
0
5
*
*
-
.
0
0
2
7
*
*
-
.
0
6
5
3
*
*
*
-
.
0
1
6
9
*
*
*
(
.
0
0
4
3
)
(
.
0
0
1
1
)
(
.
0
1
5
7
)
(
.
0
0
4
1
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
h
o
s
t
)
.
0
1
2
1
*
*
*
.
0
0
3
1
*
*
*
.
0
0
8
8
*
.
0
0
2
3
*
(
.
0
0
3
0
)
(
.
0
0
0
8
)
(
.
0
0
5
3
)
(
.
0
0
1
4
)
W
e
i
g
h
t
e
d
a

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
5
6
*
*
-
.
0
0
1
4
*
*
-
.
0
0
1
8
-
.
0
0
0
5
(
.
0
0
2
3
)
(
.
0
0
0
6
)
(
.
0
0
5
9
)
(
.
0
0
1
5
)
T
a
n
g
i
b
i
l
i
t
y
.
0
4
9
0
*
.
0
1
2
7
*
.
0
5
3
1
*
*
.
0
1
3
7
*
*
(
.
0
2
5
6
)
(
.
0
0
6
6
)
(
.
0
2
5
9
)
(
.
0
0
6
7
)
L
o
s
s
c
a
r
r
y
f
o
r
w
a
r
d
.
1
0
1
5
*
*
*
.
0
2
6
3
*
*
*
.
1
0
1
9
*
*
*
.
0
2
6
4
*
*
*
(
.
0
0
7
3
)
(
.
0
0
1
9
)
(
.
0
0
7
3
)
(
.
0
0
1
9
)
S
a
l
e
s
.
0
0
4
7
.
0
0
1
2
.
0
1
0
7
.
0
0
2
8
(
.
0
1
1
4
)
(
.
0
0
3
0
)
(
.
0
1
1
3
)
(
.
0
0
2
9
)
N
o
t
e
s
:
2
2
7
,
5
5
8
O
b
s
e
r
v
a
t
i
o
n
s
.
F
r
a
c
t
i
o
n
a
l
r
e
s
p
o
n
s
e
m
o
d
e
l
e
s
t
i
m
a
t
e
d
b
y
P
o
o
l
e
d
Q
M
L
E
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
r
e
f
e
r
s
t
o
i
n
t
e
r
n
a
l
c
r
o
s
s
-
b
o
r
d
e
r
d
e
b
t
.
A
l
l
r
e
g
r
e
s
s
i
o
n
s
i
n
c
l
u
d
e
t
i
m
e
d
u
m
m
i
e
s
a
n
d
a

l
i
a
t
e
-
s
p
e
c
i

x
e
d
e

e
c
t
s
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
.
T
h
e
s
t
a
n
d
a
r
d
e
r
r
o
r
s
a
s
s
u
m
e
c
o
u
n
t
r
y
-
y
e
a
r
c
l
u
s
t
e
r
i
n
g
a
n
d
d
i
s
r
e
g
a
r
d
r
e
p
e
a
t
e
d
o
b
s
e
r
v
a
t
i
o
n
o
f
a

l
i
a
t
e
s
o
v
e
r
t
i
m
e
(
u
n
l
i
k
e
T
a
b
l
e
s
4
-
8
)
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
A.8
T
a
b
l
e
A
.
4
:
D
e
t
e
r
m
i
n
a
n
t
s
o
f
I
n
t
e
r
n
a
l
D
e
b
t

E
x
o
g
e
n
o
u
s
a
n
d
E
n
d
o
g
e
n
o
u
s
I
n
c
e
n
t
i
v
e
s
f
o
r
S
m
a
l
l
v
s
.
L
a
r
g
e
F
i
r
m
s
E
x
o
g
.
i
n
c
e
n
t
i
v
e
s
E
n
d
o
g
.
i
n
c
e
n
t
i
v
e
s
(
c
o
m
p
.
t
o
T
a
b
l
e
4
)
(
c
o
m
p
.
t
o
T
a
b
l
e
5
)
S
m
a
l
l

r
m
s
L
a
r
g
e

r
m
s
S
m
a
l
l

r
m
s
L
a
r
g
e

r
m
s
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
h
o
s
t
)
.
8
4
3
0
*
*
*
.
2
2
8
0
*
*
*
.
6
0
4
2
*
*
*
.
1
6
2
4
*
*
*
3
.
1
1
1
1
*
*
*
.
8
4
0
7
*
*
*
2
.
8
9
1
6
*
*
*
.
7
7
5
0
*
*
*
(
.
2
5
6
6
)
(
.
0
6
9
3
)
(
.
2
1
6
1
)
(
.
0
5
7
9
)
(
.
9
0
0
7
)
(
.
2
4
2
7
)
(
.
7
2
4
2
)
(
.
1
9
4
4
)
W
e
i
g
h
t
e
d
c
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
.
2
4
2
5
.
0
6
5
6
-
.
0
7
8
6
-
.
0
2
1
1
-
2
.
3
0
2
1
*
*
-
.
6
2
2
1
*
*
-
2
.
8
8
2
9
*
*
*
-
.
7
7
2
6
*
*
*
(
.
1
7
5
5
)
(
.
0
4
7
5
)
(
.
1
0
5
8
)
(
.
0
2
8
4
)
(
.
9
7
8
2
)
(
.
2
6
3
9
)
(
.
8
8
8
0
)
(
.
2
3
8
6
)
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
h
o
s
t
)
.
0
0
0
8
*
.
0
0
0
2
*
.
0
0
0
6
.
0
0
0
2
.
0
0
0
7
.
0
0
0
2
-
.
0
0
0
2
-
.
0
0
0
0
(
.
0
0
0
5
)
(
.
0
0
0
1
)
(
.
0
0
0
4
)
(
.
0
0
0
1
)
(
.
0
0
0
9
)
(
.
0
0
0
2
)
(
.
0
0
0
8
)
(
.
0
0
0
2
)
W
e
i
g
h
t
e
d

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
0
1
-
.
0
0
0
0
-
.
0
0
0
3
-
.
0
0
0
1
-
.
0
0
0
2
-
.
0
0
0
1
.
0
0
0
7
.
0
0
0
2
(
.
0
0
0
3
)
(
.
0
0
0
1
)
(
.
0
0
0
2
)
(
.
0
0
0
1
)
(
.
0
0
1
0
)
(
.
0
0
0
3
)
(
.
0
0
1
0
)
(
.
0
0
0
2
)
I
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
h
o
s
t
)
0
.
0
6
8
1
*
.
0
1
8
4
*
.
0
2
6
1
.
0
0
7
0
.
1
0
4
3
*
*
.
0
2
8
2
*
*
.
0
7
8
5
*
.
0
2
1
0
*
(
.
0
3
7
2
)
(
.
0
1
0
0
)
(
.
0
3
6
7
)
(
.
0
0
9
8
)
(
.
0
5
1
7
)
(
.
0
1
4
0
)
(
.
0
4
7
4
)
(
.
0
1
2
7
)
W
e
i
g
h
t
e
d
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
1
8
6
*
*
-
.
0
0
5
0
*
*
-
.
0
0
9
5
-
.
0
0
2
6
-
.
0
6
9
6
-
.
0
1
8
8
-
.
0
6
9
9
*
*
-
.
0
1
8
7
*
*
(
.
0
0
7
7
)
(
.
0
0
2
1
)
(
.
0
0
6
6
)
(
.
0
0
1
8
)
(
.
0
4
2
9
)
(
.
0
1
1
6
)
(
.
0
3
5
6
)
(
.
0
0
9
5
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
h
o
s
t
)
.
0
1
0
4
*
.
0
0
2
8
*
.
0
0
8
4
*
*
*
.
0
0
2
2
*
*
*
.
0
1
0
8
.
0
0
2
9
.
0
3
8
9
*
*
*
.
0
1
0
4
*
*
*
(
.
0
0
5
7
)
(
.
0
0
1
5
)
(
.
0
0
2
8
)
(
.
0
0
0
8
)
(
.
0
1
3
0
)
(
.
0
0
3
5
)
(
.
0
0
9
4
)
(
.
0
0
2
5
)
W
e
i
g
h
t
e
d
a

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
.
0
0
5
5
.
0
0
1
5
-
.
0
0
6
6
*
*
*
-
.
0
0
1
8
*
*
*
.
0
0
4
7
.
0
0
1
3
-
.
0
4
5
6
*
*
*
-
.
0
1
2
2
*
*
*
(
.
0
0
4
4
)
(
.
0
0
1
2
)
(
.
0
0
2
6
)
(
.
0
0
0
7
)
(
.
0
1
4
2
)
(
.
0
0
3
8
)
(
.
0
1
1
9
)
(
.
0
0
3
2
)
T
a
n
g
i
b
i
l
i
t
y
.
0
5
3
7
.
0
1
4
5
-
.
0
0
7
3
-
.
0
0
2
0
.
0
5
3
2
.
0
1
4
4
-
.
0
0
3
0
-
.
0
0
0
8
(
.
0
5
0
2
)
(
.
0
1
3
6
)
(
.
0
5
3
2
)
(
.
0
1
4
3
)
(
.
0
5
0
2
)
(
.
0
1
3
6
)
(
.
0
5
3
6
)
(
.
0
1
4
4
)
L
o
s
s
c
a
r
r
y
f
o
r
w
a
r
d
.
1
3
6
8
*
*
*
.
0
3
7
0
*
*
*
.
0
8
0
8
*
*
*
.
0
2
1
7
*
*
*
1
.
1
3
5
9
*
*
*
.
0
3
6
7
*
*
*
.
0
8
1
0
*
*
*
.
0
2
1
7
*
*
*
(
.
0
1
4
0
)
(
.
0
0
3
8
)
(
.
0
1
2
4
)
(
.
0
0
3
4
)
(
.
0
1
4
2
)
(
.
0
0
3
8
)
(
.
0
1
2
7
)
(
.
0
0
3
4
)
S
a
l
e
s
-
6
.
2
5
2
6
*
*
*
-
1
.
6
9
1
2
*
*
*
-
.
0
0
4
2
-
.
0
0
1
1
-
6
.
1
0
9
6
*
*
*
-
1
.
6
5
1
0
*
*
*
.
0
0
1
9
.
0
0
0
5
(
1
.
0
9
6
6
)
(
.
2
9
6
9
)
(
.
0
1
1
9
)
(
.
0
0
3
2
)
(
1
.
0
6
6
2
)
(
.
2
8
8
5
)
(
.
0
1
1
8
)
(
.
0
0
3
2
)
O
b
s
e
r
v
a
t
i
o
n
s
5
6
,
8
8
0
5
6
,
9
0
3
5
6
,
8
8
0
5
6
,
9
0
3
N
o
t
e
s
:
F
r
a
c
t
i
o
n
a
l
r
e
s
p
o
n
s
e
m
o
d
e
l
e
s
t
i
m
a
t
e
d
b
y
p
o
o
l
e
d
Q
M
L
E
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
r
e
f
e
r
s
t
o
i
n
t
e
r
n
a
l
c
r
o
s
s
-
b
o
r
d
e
r
d
e
b
t
.
A
l
l
r
e
g
r
e
s
s
i
o
n
s
i
n
c
l
u
d
e
t
i
m
e
d
u
m
m
i
e
s
a
n
d
a

l
i
a
t
e
-
s
p
e
c
i

x
e
d
e

e
c
t
s
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
a
n
d
b
a
s
e
d
o
n
p
a
n
e
l
b
o
o
t
s
t
r
a
p
p
i
n
g
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
A.9
T
a
b
l
e
A
.
5
:
D
e
t
e
r
m
i
n
a
n
t
s
o
f
I
n
t
e
r
n
a
l
D
e
b
t

E
x
o
g
e
n
o
u
s
a
n
d
E
n
d
o
g
e
n
o
u
s
I
n
c
e
n
t
i
v
e
s
E
x
c
l
u
d
i
n
g
N
o
n
-
t
a
x
F
u
n
d
a
m
e
n
t
a
l
s
(
E
x
c
e
p
t
f
o
r
T
a
n
g
i
b
i
l
i
t
y
,
L
o
s
s
c
a
r
r
y
f
o
r
w
a
r
d
,
a
n
d
S
a
l
e
s
)
E
x
o
g
.
i
n
c
e
n
t
i
v
e
s
E
n
d
o
g
.
i
n
c
e
n
t
i
v
e
s
(
c
o
m
p
.
t
o
T
a
b
l
e
4
)
(
c
o
m
p
.
t
o
T
a
b
l
e
5
)
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
h
o
s
t
)
1
.
0
9
3
*
*
*
.
2
8
5
3
*
*
*
1
.
1
7
5
6
*
*
*
.
3
0
5
9
*
*
*
4
.
0
5
5
3
*
*
*
1
.
0
5
4
6
*
*
*
(
.
1
0
3
3
)
(
.
0
2
6
8
)
(
.
1
1
8
4
)
(
.
0
3
0
8
)
(
.
3
2
9
3
)
(
.
0
8
5
5
)
W
e
i
g
h
t
e
d
c
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
1
8
2
9
*
*
*
-
.
0
4
7
6
*
*
*
-
3
.
5
7
9
6
*
*
*
-
.
9
3
0
9
*
*
*
(
.
0
6
5
1
)
(
.
0
1
6
9
)
(
.
3
7
0
6
)
(
.
0
9
6
3
)
T
a
n
g
i
b
i
l
i
t
y
.
0
5
3
9
*
*
*
.
0
1
4
1
*
*
*
.
0
5
0
1
*
*
.
0
1
3
0
*
*
.
0
5
1
7
*
*
.
0
1
3
4
*
*
(
.
0
2
0
3
)
(
.
0
0
5
3
)
(
.
0
2
0
3
)
(
.
0
0
5
3
)
(
.
0
2
0
4
)
(
.
0
0
5
3
)
L
o
s
s
c
a
r
r
y
f
o
r
w
a
r
d
.
1
0
1
4
*
*
*
.
0
2
6
5
*
*
*
.
1
0
1
7
*
*
*
.
0
2
6
5
*
*
*
.
1
0
2
5
*
*
*
.
0
2
6
7
*
*
*
(
.
0
0
6
4
)
(
.
0
0
1
7
)
(
.
0
0
6
4
)
(
.
0
0
1
7
)
(
.
0
0
6
5
)
(
.
0
0
1
7
)
S
a
l
e
s
.
0
0
9
0
.
0
0
2
3
.
0
1
0
9
.
0
0
2
8
.
0
1
9
2
*
*
.
0
0
5
0
*
*
(
.
0
0
9
9
)
(
.
0
0
2
6
)
(
.
0
0
9
9
)
(
.
0
0
2
6
)
(
.
0
0
9
8
)
(
.
0
0
2
6
)
N
o
t
e
s
:
2
2
7
,
5
5
8
o
b
s
e
r
v
a
t
i
o
n
s
.
F
r
a
c
t
i
o
n
a
l
r
e
s
p
o
n
s
e
m
o
d
e
l
e
s
t
i
m
a
t
e
d
b
y
p
o
o
l
e
d
Q
M
L
E
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
r
e
f
e
r
s
t
o
i
n
t
e
r
n
a
l
c
r
o
s
s
-
b
o
r
d
e
r
d
e
b
t
.
A
l
l
r
e
g
r
e
s
s
i
o
n
s
i
n
c
l
u
d
e
t
i
m
e
d
u
m
m
i
e
s
a
n
d
a

l
i
a
t
e
-
s
p
e
c
i

x
e
d
e

e
c
t
s
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
a
n
d
b
a
s
e
d
o
n
p
a
n
e
l
b
o
o
t
s
t
r
a
p
p
i
n
g
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
A.10
T
a
b
l
e
A
.
6
:
D
e
t
e
r
m
i
n
a
n
t
s
o
f
I
n
t
e
r
n
a
l
D
e
b
t

E
x
o
g
e
n
o
u
s
a
n
d
E
n
d
o
g
e
n
o
u
s
I
n
c
e
n
t
i
v
e
s
U
s
i
n
g
L
i
n
e
a
r
F
i
x
e
d
E
f
f
e
c
t
s
M
o
d
e
l
s
E
x
o
g
.
i
n
c
e
n
t
i
v
e
s
E
n
d
o
g
.
i
n
c
e
n
t
i
v
e
s
(
c
o
m
p
.
t
o
T
a
b
l
e
4
)
(
c
o
m
p
.
t
o
T
a
b
l
e
5
)
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
h
o
s
t
)
.
1
5
4
9
*
*
*
-
.
1
1
9
9
(
.
0
3
0
1
)
(
.
1
2
0
5
)
W
e
i
g
h
t
e
d
c
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
1
6
9
.
3
1
0
6
*
*
(
.
0
1
8
9
)
(
.
1
4
0
9
)
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
h
o
s
t
)
.
0
0
0
0
3
.
0
0
0
0
3
(
.
0
0
0
1
)
(
.
0
0
0
1
)
W
e
i
g
h
t
e
d

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
0
1
*
*
*
-
.
0
0
0
1
(
.
0
0
0
0
2
)
(
.
0
0
0
2
)
I
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
h
o
s
t
)
.
0
0
2
1
.
0
0
2
5
(
.
0
0
4
6
)
(
.
0
0
7
8
)
W
e
i
g
h
t
e
d
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
2
4
*
*
*
-
.
0
0
1
9
(
.
0
0
0
8
)
(
.
0
0
7
5
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
h
o
s
t
)
.
0
0
3
0
*
*
*
.
0
0
8
0
*
*
*
(
.
0
0
0
5
)
(
.
0
0
1
6
)
W
e
i
g
h
t
e
d
a

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
O
P
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
1
8
*
*
*
-
.
0
0
7
9
*
*
*
(
.
0
0
0
4
)
(
.
0
0
1
9
)
T
a
n
g
i
b
i
l
i
t
y
.
0
0
5
8
.
0
0
5
2
(
.
0
0
6
2
)
(
.
0
0
6
2
)
L
o
s
s
c
a
r
r
y
f
o
r
w
a
r
d
.
0
2
7
6
*
*
*
.
0
2
7
6
*
*
*
(
.
0
0
1
6
)
(
.
0
0
1
6
)
S
a
l
e
s
.
0
0
3
7
.
0
0
2
6
(
.
0
0
3
5
)
(
.
0
0
3
6
)
N
o
t
e
s
:
2
2
7
,
5
5
8
O
b
s
e
r
v
a
t
i
o
n
s
.
L
i
n
e
a
r

x
e
d
e

e
c
t
s
m
o
d
e
l
s
(
d
i
s
r
e
g
a
r
d
i
n
g
t
h
e
b
o
u
n
d
e
d
n
e
s
s
o
f
t
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
)
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
r
e
f
e
r
s
t
o
i
n
t
e
r
n
a
l
c
r
o
s
s
-
b
o
r
d
e
r
d
e
b
t
.
T
h
e

r
s
t
c
o
l
u
m
n
r
e
p
r
e
s
e
n
t
s
a
s
t
a
n
d
a
r
d

x
e
d
e

e
c
t
s
m
o
d
e
l
w
h
i
l
e
t
h
e
s
e
c
o
n
d
c
o
l
u
m
n
r
e
p
r
e
s
e
n
t
s
a

x
e
d
e

e
c
t
s
t
w
o
-
s
t
a
g
e
l
e
a
s
t
s
q
u
a
r
e
s
m
o
d
e
l
.
D
u
e
t
o
t
h
e
i
m
p
o
s
e
d
l
i
n
e
a
r
i
t
y
,
t
h
e
c
o
e

c
i
e
n
t
s
r
e

e
c
t
p
a
r
t
i
a
l
(
a
n
d
a
v
e
r
a
g
e
p
a
r
t
i
a
l
)
e

e
c
t
s
,
u
n
l
i
k
e
i
n
T
a
b
l
e
s
4
a
n
d
5
.
A
l
l
r
e
g
r
e
s
s
i
o
n
s
i
n
c
l
u
d
e
t
i
m
e
d
u
m
m
i
e
s
a
n
d
a

l
i
a
t
e
-
s
p
e
c
i

x
e
d
e

e
c
t
s
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
A.11
Table A.7: Determinants of Internal Debt including Interaction Terms
Exogenous and Endogenous Incentives
Coe. APE Coe. APE
Corporate income tax (host):
it
.1194 .0309 63.6379*** 16.4573***
(.6526) (.1689) (17.9019) (4.6215)
Weighted corporate income tax (other locations):
w
it
-.3862 -.1000 -87.3693*** -22.5944***
(.5993) (.1549) (23.9995) (6.1957)
Financial underdevelopment (host):
it
.0022* .0006* .1044** .0270**
(.0011) (.0003) (.0486) (.0126)
Weighted nancial underdevelopment (other locations):
w
it
-.0024** -.0006** -.1511** -.0391**
(.0010) (.0003) (.0712) (.0184)
Institutional weakness (host):
it
-.0374 -.0097 .3165 .0819
(.0408) (.0105) (1.0868) (.2809)
Weighted institutional weakness (other locations):
w
it
-.0357 -.0093 -.4305 -.1113
(.0333) (.0086) (1.4465) (.3738)
Aliate-level productivity OP (host):
it
.0057 .0015 -.9164** -.2370**
(.0120) (.0031) (.3742) (.0966)
Weighted aliate-level productivity OP (other locations):
w
it
-.0189** -.0049** 1.1925** .3084**
(.0087) (.0022) (.5014) (.1295)
Interaction term:
it

it
-.0027* -.0007* -.1441*** -.0373***
(.0016) (.0004) (.0417) (.0108)
Interaction term:
w
it

w
it
.0086*** .0022*** .2081*** .0538***
(.0015) (.0004) (.0574) (.0148)
Interaction term:
it

it
.0756 .0196 .4584 .1185
(.0640) (.0165) (.9314) (.2406)
Interaction term:
w
it

w
it
-.1287** -.0333** -.6893 -.1782
(.0606) (.0157) (1.1356) (.2934)
Interaction term:
it

it
-.0406* -.0105* 1.6008*** .4140***
(.0238) (.0062) (.5496) (.1419)
Interaction term:
w
it

w
it
.0381* .0099* -2.1500*** -.5560***
(.0211) (.0054) (.7114) (.1837)
Interaction term:
it

it
.0002*** .0001*** -.0026 -.0007
(.0001) (.00002) (.0022) (.0006)
Interaction term:
w
it

w
it
.0001 .00003 .0044 .0011
(.0001) (.00002) (.0033) (.0008)
Interaction term:
it

it
.0001* .00002* .0022 .0006
(.00004) (.00001) (.0014) (.0004)
Interaction term:
w
it

w
it
.00004 -.00001 -.0032 -.0008
(.00003) (-.00001) (.0021) (.0005)
Interaction term:
it

it
.0013 .0003 -.0045 -.0012
(.0013) (.0003) (.0232) (.0060)
Interaction term:
w
it

w
it
-.0016 -.0004 .0104 .0027
(.0010) (.0003) (.0324) (.0084)
Tangibility .0502** .0130** .0209 .0054
(.0243) (.0063) (.0289) (.0075)
Loss carryforward .1010*** .0261*** .1043*** .0270***
(.0059) (.0015) (.0084) (.0022)
Sales .0058 .0015 -.0497** -.0128**
(.0114) (.0030) (.0236) (.0061)
Notes: 227,558 Observations. Fractional response model estimated by Pooled QMLE. The dependent variable refers
to internal cross-border debt. All regressions include time dummies and aliate-specic xed eects. Robust standard
errors reported in parentheses and based on panel bootstrapping. *, **, and *** indicate signicance at 10%, 5%, and
1%, respectively.
A.12
T
a
b
l
e
A
.
8
:
D
e
t
e
r
m
i
n
a
n
t
s
o
f
I
n
t
e
r
n
a
l
D
e
b
t

E
x
o
g
e
n
o
u
s
I
n
c
e
n
t
i
v
e
s
f
o
r
V
a
r
i
o
u
s
P
r
o
d
u
c
t
i
v
i
t
y
M
e
a
s
u
r
e
s
O
l
l
e
y
-
P
a
k
e
s
L
e
v
i
n
s
o
h
n
-
P
e
t
r
i
n
A
l
t
e
r
n
a
t
i
v
e
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
h
o
s
t
)
1
.
0
0
1
0
*
*
*
.
2
6
1
5
*
*
*
1
.
0
2
1
4
*
*
*
.
2
6
5
1
*
*
*
1
.
0
5
8
5
*
*
*
.
2
7
3
8
*
*
*
(
.
1
1
8
1
)
(
.
0
3
0
7
)
(
.
1
1
5
0
)
(
.
0
2
9
9
)
(
.
1
1
9
4
)
(
.
0
3
1
0
)
W
e
i
g
h
t
e
d
c
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
3
6
8
-
.
0
0
9
5
-
.
0
5
7
8
-
.
0
1
5
0
-
.
0
5
4
6
-
.
0
1
4
1
(
.
0
0
0
2
)
(
.
0
1
6
4
(
.
0
6
3
7
)
(
.
0
1
6
5
)
(
.
0
7
1
7
)
(
.
0
1
8
5
)
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
h
o
s
t
)
.
0
0
0
8
*
*
*
.
0
0
0
2
*
*
*
.
0
0
0
8
*
*
*
.
0
0
0
2
*
*
*
.
0
0
0
9
*
*
*
.
0
0
0
2
*
*
*
(
.
0
0
0
2
)
(
.
0
0
0
0
)
(
.
0
0
0
2
)
(
.
0
0
0
0
)
(
.
0
0
0
2
)
(
.
0
0
0
1
)
W
e
i
g
h
t
e
d

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
0
4
*
*
*
-
.
0
0
0
1
*
*
*
-
.
0
0
0
4
*
*
*
-
.
0
0
0
1
*
*
*
-
.
0
0
0
4
*
*
*
-
.
0
0
0
1
*
*
*
(
.
0
0
0
1
)
(
.
0
0
0
0
)
(
.
0
0
0
1
)
(
.
0
0
0
0
)
(
.
0
0
0
1
)
(
.
0
0
0
0
)
I
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
h
o
s
t
)
.
0
2
2
1
.
0
0
5
7
.
0
2
0
6
.
0
0
5
4
.
0
2
0
4
.
0
0
5
3
(
.
0
1
7
7
)
(
.
0
0
4
6
)
(
.
0
1
6
8
)
(
.
0
0
4
4
)
(
.
0
2
0
3
)
(
.
0
0
5
3
)
W
e
i
g
h
t
e
d
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
1
0
6
*
*
*
-
.
0
0
2
7
*
*
*
-
.
0
1
2
0
*
*
*
-
.
0
0
3
1
*
*
*
-
.
0
1
1
7
*
*
*
-
.
0
0
3
0
*
*
*
(
.
0
0
3
2
)
(
.
0
0
0
8
)
(
.
0
0
3
2
)
(
.
0
0
0
8
)
(
.
0
0
3
3
)
(
.
0
0
0
9
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
(
a
v
g
.
w
i
t
h
i
n
h
o
s
t
)
.
0
1
1
7
*
*
*
.
0
0
3
0
*
*
*
.
0
0
0
0
*
*
*
.
0
0
0
0
*
*
*
.
0
0
0
0
.
0
0
0
0
(
.
0
0
2
2
)
(
.
0
0
0
6
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
(
d
e
v
.
f
r
o
m
a
v
g
.
w
i
t
h
i
n
h
o
s
t
)
.
0
1
2
5
*
*
*
.
0
0
3
2
*
*
*
.
0
0
0
0
.
0
0
0
0
.
0
0
0
0
*
*
*
.
0
0
0
0
*
*
*
(
.
0
0
1
9
)
(
.
0
0
0
5
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
W
e
i
g
h
t
e
d
a

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
5
5
*
*
*
-
.
0
0
1
4
*
*
*
-
.
0
0
0
0
*
*
*
-
.
0
0
0
0
*
*
*
-
.
0
0
0
0
*
*
-
.
0
0
0
0
*
*
(
.
0
0
1
5
)
(
.
0
0
0
4
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
T
a
n
g
i
b
i
l
i
t
y
.
0
4
9
0
*
*
.
0
1
2
7
*
*
.
0
4
5
0
*
.
0
1
1
7
*
.
0
4
5
5
*
.
0
1
1
8
*
(
.
0
2
4
2
)
(
.
0
0
6
3
)
(
.
0
2
5
1
)
(
.
0
0
6
5
)
(
.
0
2
5
0
)
(
.
0
0
6
5
)
L
o
s
s
c
a
r
r
y
f
o
r
w
a
r
d
.
1
0
1
5
*
*
*
.
0
2
6
3
*
*
*
.
1
0
2
3
*
*
*
.
0
2
6
6
*
*
*
.
1
0
2
4
*
*
*
.
0
2
6
5
*
*
*
(
.
0
0
5
9
)
(
.
0
0
1
5
)
(
.
0
0
5
7
)
(
.
0
0
1
5
)
(
.
0
0
5
7
)
(
.
0
0
1
5
)
S
a
l
e
s
.
0
0
4
5
.
0
0
1
1
.
0
1
1
3
.
0
0
2
9
.
0
1
1
6
.
2
6
1
5
(
.
0
1
1
4
)
(
.
0
0
3
0
)
(
.
0
1
1
0
)
(
.
0
0
2
8
)
(
.
0
1
0
7
)
(
.
0
0
3
0
)
N
o
t
e
s
:
2
2
7
,
5
5
8
O
b
s
e
r
v
a
t
i
o
n
s
.
F
r
a
c
t
i
o
n
a
l
r
e
s
p
o
n
s
e
m
o
d
e
l
e
s
t
i
m
a
t
e
d
b
y
P
o
o
l
e
d
Q
M
L
E
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
r
e
f
e
r
s
t
o
i
n
t
e
r
n
a
l
c
r
o
s
s
-
b
o
r
d
e
r
d
e
b
t
.
A
l
l
r
e
g
r
e
s
s
i
o
n
s
i
n
c
l
u
d
e
t
i
m
e
d
u
m
m
i
e
s
a
n
d
a

l
i
a
t
e
-
s
p
e
c
i

x
e
d
e

e
c
t
s
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
a
n
d
b
a
s
e
d
o
n
p
a
n
e
l
b
o
o
t
s
t
r
a
p
p
i
n
g
.
T
h
e

r
s
t
t
w
o
c
o
l
u
m
n
s
s
h
o
u
l
d
b
e
c
o
m
p
a
r
e
d
w
i
t
h
T
a
b
l
e
4
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
A.13
T
a
b
l
e
A
.
9
:
D
e
t
e
r
m
i
n
a
n
t
s
o
f
I
n
t
e
r
n
a
l
D
e
b
t

E
n
d
o
g
e
n
o
u
s
I
n
c
e
n
t
i
v
e
s
f
o
r
V
a
r
i
o
u
s
P
r
o
d
u
c
t
i
v
i
t
y
M
e
a
-
s
u
r
e
s
O
l
l
e
y
-
P
a
k
e
s
L
e
v
i
n
s
o
h
n
-
P
e
t
r
i
n
A
l
t
e
r
n
a
t
i
v
e
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
e

.
A
P
E
C
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
h
o
s
t
)
3
.
5
1
7
3
*
*
*
.
9
0
9
6
*
*
*
3
.
3
4
9
0
*
*
*
.
8
6
8
4
*
*
*
3
.
3
5
1
5
*
*
*
.
8
6
6
1
*
*
*
(
.
4
7
3
9
)
(
.
1
2
2
5
)
(
.
4
1
8
9
)
(
.
1
0
8
3
)
(
.
4
4
3
9
)
(
.
1
1
4
5
)
W
e
i
g
h
t
e
d
c
o
r
p
o
r
a
t
e
i
n
c
o
m
e
t
a
x
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
2
.
9
4
3
1
*
*
*
-
.
7
6
1
1
*
*
*
-
2
.
7
5
6
6
*
*
*
-
.
7
1
4
8
*
*
*
-
2
.
7
1
3
3
*
*
*
-
.
7
0
1
2
*
*
*
(
.
5
3
9
6
)
(
.
1
3
9
4
)
(
.
4
7
9
0
)
(
.
1
2
3
8
)
(
.
5
2
5
8
)
(
.
1
3
5
7
)
F
i
n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
h
o
s
t
)
.
0
0
1
0
*
*
.
0
0
0
2
*
*
.
0
0
0
8
*
.
0
0
0
2
*
.
0
0
0
8
*
.
0
0
0
2
*
(
.
0
0
0
4
)
(
.
0
0
0
1
)
(
.
0
0
0
4
)
(
.
0
0
0
1
)
(
.
0
0
0
4
)
(
.
0
0
0
1
)
W
e
i
g
h
t
e
d

n
a
n
c
i
a
l
u
n
d
e
r
d
e
v
e
l
o
p
m
e
n
t
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
0
7
-
.
0
0
0
2
-
.
0
0
0
5
-
.
0
0
0
1
-
.
0
0
0
4
-
.
0
0
0
1
(
.
0
0
0
5
)
(
.
0
0
0
1
)
(
.
0
0
0
5
)
(
.
0
0
0
1
)
(
.
0
0
0
5
)
(
.
0
0
0
1
)
I
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
h
o
s
t
)
.
0
5
8
8
*
*
.
1
5
2
1
*
*
.
0
6
5
1
*
*
*
.
0
1
6
9
*
*
*
.
0
6
9
8
*
*
.
0
1
8
0
*
*
(
.
0
2
7
6
)
(
.
0
0
7
1
)
(
.
0
2
4
5
)
(
.
0
0
6
4
)
(
.
0
2
8
5
)
(
.
0
0
7
4
)
W
e
i
g
h
t
e
d
i
n
s
t
i
t
u
t
i
o
n
a
l
w
e
a
k
n
e
s
s
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
6
0
3
*
*
*
-
.
0
1
5
6
*
*
*
-
.
0
7
0
6
*
*
*
-
.
0
1
8
3
*
*
*
-
.
0
7
5
3
*
*
*
-
.
0
1
9
5
*
*
*
(
.
0
2
3
4
)
(
.
0
0
6
0
)
(
.
0
2
0
5
)
(
.
0
0
5
3
)
(
.
0
2
3
6
)
(
.
0
0
6
1
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
(
a
v
g
.
w
i
t
h
i
n
h
o
s
t
)
.
0
1
4
9
*
.
0
0
3
9
*
-
.
0
0
0
0
-
.
0
0
0
0
.
0
0
0
0
.
0
0
0
0
(
.
0
0
8
0
)
(
.
0
0
2
1
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
(
.
0
0
0
1
)
(
.
0
0
0
0
)
A

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
(
d
e
v
.
f
r
o
m
a
v
g
.
w
i
t
h
i
n
h
o
s
t
)
.
0
1
6
0
*
*
*
.
0
0
4
1
*
*
*
-
.
0
0
0
0
-
.
0
0
0
0
.
0
0
0
0
.
0
0
0
0
(
.
0
0
5
9
)
(
.
0
0
1
5
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
(
.
0
0
0
1
)
(
.
0
0
0
0
)
W
e
i
g
h
t
e
d
a

l
i
a
t
e
-
l
e
v
e
l
p
r
o
d
u
c
t
i
v
i
t
y
(
o
t
h
e
r
l
o
c
a
t
i
o
n
s
)
-
.
0
0
9
8
-
.
0
0
2
5
.
0
0
0
0
.
0
0
0
0
-
.
0
0
0
0
.
0
0
0
0
(
.
0
0
8
6
)
(
.
0
0
2
2
)
(
.
0
0
0
0
)
(
.
0
0
0
0
)
(
.
0
0
0
1
)
(
.
0
0
0
0
)
T
a
n
g
i
b
i
l
i
t
y
.
0
5
2
3
*
*
.
0
1
3
5
*
*
.
0
5
1
5
*
*
.
0
1
3
4
*
*
.
0
4
8
9
*
.
0
1
2
6
*
(
.
0
2
4
1
)
(
.
0
0
6
2
)
(
.
0
2
5
2
)
(
.
0
0
6
5
)
(
.
0
2
5
1
)
(
.
0
0
6
5
)
L
o
s
s
c
a
r
r
y
f
o
r
w
a
r
d
.
1
0
2
0
*
*
*
.
0
2
6
4
*
*
*
.
1
0
2
7
*
*
*
.
0
2
6
6
*
*
*
.
1
0
2
9
*
*
*
.
0
2
6
6
*
*
*
(
.
0
0
6
0
)
(
.
0
0
1
5
)
(
.
0
0
5
8
)
(
.
0
0
1
5
)
(
.
0
0
5
8
)
(
.
0
0
1
5
)
S
a
l
e
s
.
0
1
0
1
.
0
0
2
6
.
0
1
7
5
.
0
0
4
5
.
0
1
6
9
.
0
0
4
3
(
.
0
1
1
7
)
(
.
0
0
3
0
)
(
.
0
1
1
3
)
(
.
0
0
2
9
)
(
.
0
1
0
7
)
(
.
0
0
2
8
)
N
o
t
e
s
:
2
2
7
,
5
5
8
O
b
s
e
r
v
a
t
i
o
n
s
.
F
r
a
c
t
i
o
n
a
l
r
e
s
p
o
n
s
e
m
o
d
e
l
e
s
t
i
m
a
t
e
d
b
y
P
o
o
l
e
d
Q
M
L
E
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
r
e
f
e
r
s
t
o
i
n
t
e
r
n
a
l
c
r
o
s
s
-
b
o
r
d
e
r
d
e
b
t
.
A
l
l
r
e
g
r
e
s
s
i
o
n
s
i
n
c
l
u
d
e
t
i
m
e
d
u
m
m
i
e
s
a
n
d
a

l
i
a
t
e
-
s
p
e
c
i

x
e
d
e

e
c
t
s
.
R
o
b
u
s
t
s
t
a
n
d
a
r
d
e
r
r
o
r
s
r
e
p
o
r
t
e
d
i
n
p
a
r
e
n
t
h
e
s
e
s
a
n
d
b
a
s
e
d
o
n
p
a
n
e
l
b
o
o
t
s
t
r
a
p
p
i
n
g
.
T
h
e

r
s
t
t
w
o
c
o
l
u
m
n
s
s
h
o
u
l
d
b
e
c
o
m
p
a
r
e
d
w
i
t
h
T
a
b
l
e
4
a
n
d
t
h
e
l
a
s
t
f
o
u
r
c
o
l
u
m
n
s
a
r
e
c
o
u
n
t
e
r
p
a
r
t
s
t
o
T
a
b
l
e
8
i
n
t
h
e
p
a
p
e
r
.
*
,
*
*
,
a
n
d
*
*
*
i
n
d
i
c
a
t
e
s
i
g
n
i

c
a
n
c
e
a
t
1
0
%
,
5
%
,
a
n
d
1
%
,
r
e
s
p
e
c
t
i
v
e
l
y
.
A.14

Похожие интересы